- Email: [email protected]
MRI assessment of surrounding tissues in soft-tissue sarcoma during neoadjuvant chemotherapy can help predicting response and prognosis
Accepted Manuscript Title: MRI Assessment of Surrounding Tissues in Soft-tissue Sarcoma During Neoadjuvant Chemotherapy Can Help Predicting Response and Prognosis Authors: Amandine Crombe, Franc¸ois Le Loarer, Eberhard Stoeckle, Sophie Cousin, Audrey Michot, Antoine Italiano, Xavier Buy, Mich`ele Kind PII: DOI: Reference:
S0720-048X(18)30399-1 https://doi.org/10.1016/j.ejrad.2018.11.004 EURR 8364
To appear in:
European Journal of Radiology
Received date: Revised date: Accepted date:
24 June 2018 3 October 2018 4 November 2018
Please cite this article as: Crombe A, Le Loarer F, Stoeckle E, Cousin S, Michot A, Italiano A, Buy X, Kind M, MRI Assessment of Surrounding Tissues in Soft-tissue Sarcoma During Neoadjuvant Chemotherapy Can Help Predicting Response and Prognosis, European Journal of Radiology (2018), https://doi.org/10.1016/j.ejrad.2018.11.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
MRI Assessment of Surrounding Tissues in Soft-tissue Sarcoma During Neoadjuvant Chemotherapy Can Help Predicting Response and Prognosis Amandine CROMBE1,2,3, François LE LOARER3,4, Eberhard STOECKLE5, Sophie COUSIN6, Audrey MICHOT5, Antoine ITALIANO6, Xavier BUY1, Michèle KIND1.
IP T
1. Department of Radiology, Institut Bergonie, F-33000 Bordeaux, France 2. INRIA Bordeaux-Sud-Ouest, CNRS UMR 5251, F-33405 Talence, France 3. Université de Bordeaux, F-33000, Bordeaux, France
SC R
4. Department of Pathology, Institut Bergonie, F-33000 Bordeaux, France 5. Department of Surgery, Institut Bergonie, F-33000 Bordeaux, France
U
6. Department of Medical Oncology, Institut Bergonie, F-33000 Bordeaux, France
Corresponding author: Dr Amandine Crombé
N
email: [email protected]
A
fax: +33 (0) 5 56 33 33 30
M
tel: +33 (0) 5 56 33 33 33
address: Department of Radiology, Institut Bergonié
HIGHLIGHTS
Changes in surrounding tissues of STS during chemotherapy are associated with
EP
TE D
229 cours de l’Argonne, 33000 Bordeaux, France
response.
Infiltrative growth pattern and stable/increased oedema may predict worse outcome.
CC
Evaluation of surrounding tissues may help anticipating persistence of satellite
A
tumorous cells
ABSTRACT Objectives: To determine if changes in surrounding tissues of soft-tissue sarcomas (STS) evaluated by MRI during neoadjuvant chemotherapy (NAC) are associated with the histological response and satellite tumorous cells beyond the pseudocapsule on surgical specimen, disease-free survival (DFS) and overall survival (OS).
2 Methods: Fifty-seven adult patients with locally advanced high-grade STS of extremities and trunk wall were included in this single-centre retrospective study. All were uniformly treated by 5-6 cycles of anthracycline-based NAC, curative surgery and adjuvant radiotherapy and had available MRI with a gadolinium-chelates administration at baseline and after 2 cycles. Thirty-seven patients also had a pre-operative MRI. Two senior radiologists evaluated MRI growth pattern, oedema, contrast-enhanced oedema, aponeurotic enhancement, and their qualitative changes during NAC. An expert pathologist reviewed all
IP T
surgical specimens. A good histological response was defined as <10% viable cells.
Associations with pathological findings were estimated with Fisher and Chi-square tests and
SC R
multivariate analysis with binary logistic regression. Survival analyses included log-rank tests.
Results: Forty-two patients had poor responses and 25 had satellite tumorous cells on surgical specimen. Changes in surrounding oedema and in contrast-enhanced oedema were
U
associated with responses (p=0.008 and 0.011, respectively). Diffuse infiltrative growth
N
pattern (IGP) on baseline MRI, changes in margin definition and in contrast-enhanced
A
oedema at early evaluation were associated with satellite tumorous cells (p=0.039, 0.011 and 0.009, respectively). Diffuse IGP on baseline MRI and stable or increased oedema during
M
treatment were predictors of DFS (p=0.009 and 0.040, respectively) Conclusion: Surrounding tissues of STS during NAC should be carefully evaluated as they
TE D
may steer treatment efficacy and patient prognosis.
ABBREVIATIONS
EP
CI95%: 95% confidence interval
DCE-MRI: dynamic contrast enhanced MRI DFS: disease free survival
CC
DWI: diffusion weighted imaging FNCLCC: French ‘Federation Nationale des Centres de Lutte Contre le Cancer’
A
HR: Hazard ratio IGP: infiltrative growth pattern LD: longest diameter MPNST: malignant peripheral nerve sheath tumour M/RC-LPS: myxoid/round cells liposarcoma NAC: neoadjuvant chemotherapy OR: Odds ratio
3 OS: overall survival STS: soft tissue sarcoma WI: weighted imaging
KEYWORDS Soft-tissue sarcoma ; Peritumoral tissue ;
IP T
Infiltrative growth pattern ; Chemotherapy ;
SC R
Tumour response evaluation;
1. INTRODUCTION
Soft-tissue sarcomas (STS) represent a heterogeneous group of ubiquitous mesenchymal
U
tumours. About 4000 new cases of STS are annually diagnosed in France, making these
N
tumours the most frequent of rare tumours [1]. On MRI, their periphery can demonstrate a
A
wide range of anomalies. Oedema in surrounding tissue is frequently observed and might be related to high-grade tumours and to isolated or clusters of satellite tumorous cells beyond
M
the tumour borders [2-4]. Diffuse infiltrative growth pattern (IGP) on MRI, which consists in poorly defined margins, has also been associated with high-grade STS and with
TE D
histological IGP - an independent predictor of poorer prognosis [5-8]. Nakamura et al. also recently highlighted that diffuse IGP on MRI was predictive of disease–free survival (DFS) and metastasis-free survival [9]. Additionally, the ‘tail sign’, which corresponds to a thick
EP
aponeurotic enhancement after Gadolinium-chelates injection due to tumour spreading along fascia in myxofibrosarcomas and undifferentiated pleomorphic sarcoma, should be carefully removed during surgery in order to decrease the risk of local relapse [10-12].
CC
Standard of care for high-grade STS is on the verge of being redefined since anthracyclinebased neoadjuvant chemotherapy (NAC) has shown promising results on outcome with
A
more conservative surgical procedures [13-16]. Imaging evaluation of responses to treatment remains based on RECIST 1.1 despite its limited ability to predict histological response [17]. Imaging markers, such as modified Choi criteria, DCE-MRI, 18FDG/PET-CT scan, are superior to RECIST 1.1 but none has so far been used to assess tumour periphery and their therapy-related changes [17-24]. A single sub-study of 6 cases has previously linked surrounding oedema to poorer responses to chemotherapy [25]. Moreover, systematic radio-
4 pathological correlations within tumour periphery might help to adjust the surgical margins and the radiation fields for post-operative radiotherapy [26]. Hence, we aimed at (i) characterizing STS periphery with MRI at baseline and changes during NAC, (ii) correlating the MRI findings with histological response and persistence of viable satellite tumorous cells on the surgical specimen and survival, and (iii) identifying links between these MRI findings and patient outcome.
IP T
2. METHODS 2.1. Population and design of the study
SC R
Our institutional review board approved this retrospective single centre study and informed consent was waived. Between January 2008 and June 2017, we included consecutive patients treated at our institution as they presented with newly diagnosed biopsy-proven high-grade STS of trunk wall or extremities, without metastasis (assessed on whole body
U
MRI for myxoid/round cells liposarcoma and chest CT-scan for other histotypes), uniformly
N
treated with 5-6 cycles of anthracycline-based NAC followed by a curative surgery and an
A
adjuvant radiotherapy, with available baseline MRI (MRI-0) and MRI after 2-3 cycles (MRI-1). Grade was defined according to the French Fédération Nationale des Centres de
M
Lutte Contre le Cancer (FNCLCC) grading system and high-grade corresponded to grade III tumours [27].
TE D
Following co-variables were reported: age, gender, histotype, tumour location, depth relative to superficial fascia, delay from last cycle to surgery, surgical margins, adjuvant radiotherapy, occurrence of relapse (local or metastatic) or tumour-related death. At our
EP
institution, routine follow-up consists in clinical examination and chest radiograph every 3 months for 2 years, every 6 months for 5 years and then annually until 10 years after surgery. These examinations were complemented by chest CT scans (with 1mm thick
CC
reconstruction) and MRI for local evaluation in case of abnormal or doubtful findings. All
A
local or distant relapses were histopathologically confirmed.
2.2. MRI analysis We only included MRI with at least two orthogonal acquisition plans and following available sequences: (i) T2-weighted-imaging (-WI), (ii) fat suppressed T2-WI or proton density sequences and (iii) T1-WI after Gadolinium-chelates injection. All techniques to suppress fat were accepted: short TI inversion recovery, Fat-sat, Dixon methods. MRI were all performed on 1.5T MR system, mostly on the same system (Magnetom AERA 1.5T,
5 Siemens Healthineers, Erlangen, Germany) except for 30 baseline MRI that were performed in external radiological centre before the patient had been addressed to our sarcoma reference centre. Details for conventional MRI protocol performed at our institution are given in Supplementary Data. MRI analysis was performed by 2 senior radiologists from a sarcoma reference centre (with 4 and 30 years of experience in musculoskeletal tumours, respectively), by using a picture archiving and communication system workstation (Entreprise Imaging, AGFA, the
IP T
Netherlands). Each radiologist independently read the whole sets of MRI blinded to
pathological and radiological reports. A final consensual lecture was made between the two
SC R
radiologists in case of disagreement.
All patients had baseline MRI-0 (n=57), MRI-1 (n=57). Additionally, 37 patients had a preoperative MRI, at the end of NAC (called MRI-2).
The following features depicting tumour periphery were reported (Fig. 1):
U
- Growth pattern at baseline on whole tumour circumference on T2-WI and T1-WI after
N
gadolinium-chelates injection and fat suppression. It was divided in 3 categories according to Nakamura et al. [9]: ‘pushing-type’ (i.e. when the tumour was entirely well-defined),
A
‘focal-type’ (i.e. when infiltrative pattern, namely irregular borders and infiltration of
tumour circumference).
M
surrounding tissue represented <25% of tumour circumference) and ‘diffuse-type’ (≥ 25% of
TE D
- Bone, vessel and/or nerves invasion at baseline - Presence of surrounding oedema at baseline and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2 (absent or decrease versus stable or increase).
EP
Oedema was defined as ill-defined non-fatty area of bright SI on T2-WI or proton density WI beyond tumour borders, without mass effect. - Contrast uptake of the surrounding oedema (referred as ‘contrast-enhanced oedema’ in this
CC
study) at baseline and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2. Similarly, qualitative assessment was dichotomised as: absent or decrease versus
A
stable or increase. - Aponeurotic enhancement of more than 2mm thickness at baseline (referred as ‘tail sign’ in the literature [10-12] and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2, evaluated on T1-WI after Gadolinium-chelates injection and fat suppression. The radiologists also reported longest diameters at MRI-0, MRI-1 and MRI-2 (tumour-LD, in mm) and relative change from MRI-0 to MRI-1 and from MRI-0 to MRI-2 (in %), in
6 order to assess response status according to RECIST 1.1 [28]. Measures were performed on T2-WI, helped by all the other available MR sequences.
2.3. Pathological analysis Haematoxylin and eosinophil stained (HES) slices of all surgical specimens were retrospectively reviewed for the study by an expert pathologist from our sarcoma reference center, blinded to clinical course. The pathologist reported:
IP T
- The histological response with estimation of the percentage of viable tumour (stainable
cells), necrotic and post-therapy fibrosis on the entire tumour volume. A good response was
SC R
defined with the threshold of <10% viable cells, as it has been demonstrated to be predictive of worse metastatic-free survival and OS [29].
- Presence of satellite tumorous cells beyond tumour borders and pseudocapsule if present. Of note, 4 surgical specimens had not been sampled according to guidelines thus not
U
enabling an assessment of the whole periphery of the tumours. These cases were
N
consequently removed from statistical analyses focusing on the presence of satellite
A
tumorous cells as covariables or outcome. Surgical margins were classified by consensus between surgeon and pathologist. They were classified as R0 (i.e. macroscopically complete
M
with negative microscopic margins), R1 (i.e. macroscopically complete with positive
TE D
microscopic margin) and R2 (i.e. macroscopically incomplete).
2.4. Statistical analysis
Inter-observer agreements were assessed using the Cohen’s Kappa (κ) test for dichotomised variables. Weighted Kappa statistics were used for ordinal variables. The agreement for
EP
classical κ and κw was defined as slight (0-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80) and almost perfect (0.81-0.99) [30].
CC
Associations between categorical (or ordinal variables) and histological findings were performed with the Chi-2 test and Fisher test, as appropriate. A binary multivariate logistic
A
regression modelling was built to identify the MRI features that remained predictive of the histological findings. Only features with a p-value of less than 0.05 were included in the multivariate analysis. Patients with missing values were removed from the multivariate analysis. We added the following dichotomized covariables for adjustment in the multivariate analysis: histotype (undifferentiated sarcoma, which corresponded to undifferentiated pleomorphic sarcomas and myxofibrosarcomas, versus others), age (< or ≥ median age) and size (< or ≥ median size). A multivariate analysis including several
7 radiological parameters altogether was not carried out because of the limited population size. Univariate and multivariate odds ratio (OR) are given with their 95% confidence interval. Survival analysis included the overall survival (OS) and disease-free survival (DFS) in months since the curative surgery was performed. Local recurrence-free survival and metastasis-free survival were combined because of the limited population size and the small number of events. Univariate analyses were made with Kaplan-Meier log-rank tests. The hazard ratio (HR) with CI95% was calculated to measure the degree of survival differences
IP T
on the Kaplan-Meier plots. We excluded patients without any event during a follow-up of less than 2 years after curative surgery.
SC R
All tests were two-tailed. Statistical analyses were done using the SPSS statistical package (IBM, version 21.0, Chicago, IL). Variables are expressed as mean, standard deviation,
U
median and range, as appropriate. A p-value < 0.05 was deemed significant.
N
3. RESULTS
A
3.1. Population (Table 1)
Fifty-seven patients met all the inclusion criteria (21 females, median age 58 years old,
M
range: 31-77) among the 283 patients with newly diagnosed sarcoma who received NAC at our institution during the study period. Twenty-two (38.6%) had an undifferentiated
TE D
pleomorphic sarcoma, 5 (8.8%) had a myxofibrosarcoma, 8 (14%) had a rhabdomyosarcoma, 5 (8.8%) had a leiomyosarcoma, 3 (5.3%) had a pleomorphic liposarcoma, 1 (1.8%) had a dedifferentiated liposarcomas, 6 (10.5%) had myxoid/round cell
EP
liposarcomas, 6 (10.5%) had a synovial sarcoma and one presented with a malignant peripheral nerve sheath tumour (MPNST). The median size before treatment was 102 mm (range: 39–260). Only 2 were superficially situated (3.5%). Twenty patients did not have a
CC
pre-operative MRI for late evaluation.
A
3.2. Inter-observer agreements of the MRI features (Table 2) Inter-observer agreements for MRI features at baseline, early and late evaluations were all statistically significant except for the qualitative change in aponeurotic enhancement between MRI-0 and MRI-2 (κ=0.211, p=0.183). At baseline, all features depicting tumour periphery were substantially reproducible (κ>0.600) except for growth pattern, which was moderately reproducible (κ=0.427). At early evaluation, change in oedema and contrastenhanced oedema were moderately reproducible (κ=0.527 and 0.553, respectively). Change
8 in margin definition was also fairly reproducible at both early and late evaluations (κ=0.335 and 0.243, respectively).
3.3. Association with pathological findings Association with histological response (Table 3) Forty-two (73.7%) patients were poor responders. None of the baseline MRI findings was associated with the response.
IP T
Stability or increase of a pre-existing oedema at early evaluation was seen in 40% (6/15) good responders and 73.2% (30/41) in poor responders. It was associated with a poor
SC R
response at univariate analysis (OR=4.09, p=0.030).
Stability or increase of a pre-existing contrast-enhanced oedema at early evaluation was seen in 26.7% (4/15) good responders and 64.3% (27/42) poor responders, being associated of a poor response at univariate analysis (OR=4.95, p=0.017).
U
At multivariate analysis, both stable or increased oedema and contrast-enhanced oedema at
N
early evaluation remained predictors of a poor response after adjustment for age, size and
A
histotypes (OR=6.87, p=0.011 and OR=8.06, p=0.008, respectively - Table 4) Patients were mostly classified as stable disease according to RECIST 1.1: 86.7% (13/15)
M
good responders and 88.1% (37/42) poor responders at early evaluation, and 66.7% (6/9) good responders and 78.6% (22/28) poor responders at late evaluation. RECIST 1.1 did not
respectively).
TE D
correlate with histological response at early and late evaluation (p=0.710 and 0.153,
Association with satellite tumorous cells beyond tumour border
EP
Twenty-five out of 53 (47.2%) patients had viable satellite tumorous cells beyond tumour borders on surgical specimen. MRI growth pattern at baseline was associated with satellite tumorous cells (p=0.022). Seventeen out of 25 (68%) patients with satellite tumorous cells
CC
and 10 out of 28 (35.7%) patients without had diffuse infiltrative growth pattern on MRI. At early evaluation, stable or worst margin definition was associated with satellite tumorous
A
cells (OR=15.53, p=0.003). This feature was seen in 96% (24/25) of patients with satellite tumorous cells on surgical specimen and 60.7%, (17/28) of patients without. It remained significantly associated at late evaluation (OR=10.00, p=0.010). Additionally, 76% (19/25) of patients with satellite tumorous cells and 35.7% (10/28) of patients without had a stable or increased contrast-enhanced oedema at early evaluation. This MRI feature was associated with satellite tumorous cells at early evaluation (OR=5.70, p=0.005) and also at late evaluation (OR=5.42, p=0.037).
9 At multivariate analysis, diffuse IGP on baseline MRI, stability or worsening of margin definition, and stable or increased contrast-enhanced oedema remained associated with satellite tumorous cells on surgical specimen after adjustment for age, size and histotypes (OR=3.52, p=0.039; OR=19.06, p=0.011, and OR=5.59, p=0.009, respectively – Table 4).
3.4. Prognostic value of radio-pathological findings involving surrounding tissue (Table 5) Survival analysis was carried on 31 patients. The mean survival of the series was 76 months
IP T
(CI95%=(56-97)). Thirteen out of 31 patients (41.9%) had died of the disease and one died of other causes, 19 out of 31 patients (61.3%) had a relapse, either local (7/31, 22.6%) or
SC R
distant (17/31, 54.8%), mostly lung metastases. None were lost to follow-up. Four of them
were not evaluable for satellite tumorous cells because of inadequate and limited sampling of tumor periphery. Only 19 (52.8%, 19 out of 36) had available MRI-2 for late evaluation.
U
At baseline and early evaluation, diffuse IGP and stable or increased oedema were predictors
N
of DFS (HR=3.72, p=0.009 and HR=2.80, p=0.040, respectively).
A
At late evaluation, stable or increased oedema was significantly associated with a lower OS and DFS (HR=5.69, p=0.019 and HR=4.98, p=0.006, respectively), as well as change in
M
aponeurotic enhancement (HR=5.77, p=0.012 and HR=6.06, p=0.003, respectively). Interestingly, presence of satellite tumorous cells on surgical specimen was predictive of
TE D
lower OS and DFS at univariate analysis (HR=10.47 (1.35-81.5), p=0.005 and HR=4.64 (1.30-16.5), p=0.010, respectively – Fig. 2).
EP
3.5. Association between histological infiltrative pattern and histological response Good histological response and absence of satellite tumorous cells were not systematically associated as 4 good responders demonstrated satellite tumorous cells on surgical specimen.
CC
Three out of these 4 patients showed local (2 out of 4) and/or distant relapse in lung (2 out of 4) during follow-up and two of them died of disease 17 and 44.5 months after surgery.
A
Figure 3 illustrates one of these 4 cases. Figure 4 illustrate an opposite case of a living good responder without satellite tumorous cells with complete regression of peripheral anomalies on MRI. We also identified 4 patients with good responses and no satellite tumorous cells who demonstrated stable or slight increase of surrounding oedema and contrast-enhanced oedema. Radio-pathological correlations demonstrated a halo of inflammatory cells and activated myofibroblasts around a fibrotic capsule (Fig. 5).
4. DISCUSSION Surrounding tissues of STS demonstrate various aspects at baseline and during treatments but their relevance to predict histological response and prognosis is poorly understood. In addition, a better characterization of viable satellite tumorous cells beyond apparent tumorous borders is needed to delineate accurately surgical margins and radiotherapy fields. Overall, our results indicate that diffuse IGP on baseline MRI, poorly defined margins at early and late evaluation, as well as stable or increased contrast-enhanced oedema could help
IP T
to predict satellite tumorous cells on surgical specimen. Furthermore, changes in
surrounding oedema and in enhancement of this oedema may help to anticipate histological
SC R
response. Finally, diffuse IGP on baseline MRI and early change in surrounding oedema may predict of DFS.
Our results confirm those observed by Hanna et al. [25], that oedema would be associated with responses to NAC. Our results at baseline are also in accordance with studies that
U
showed associations between MRI IGP and histological IGP, as well as higher risk of poorer
N
DFS with diffuse IGP on MRI [7, 9]. Our results deepen these findings by investigating the
A
predictive value of early and late changes in surrounding tissues during NAC. In addition, we observed that therapy-related changes may dissociate with a good response within
M
tumour bulk while viable tumorous cells still persist in tissue periphery. Most patients of our cohort with this pattern of response displayed distant and local recurrences (3 of 4). We
TE D
hypothesize that tumour periphery represent in such cases the active invasion front of tumours, providing important insights regarding recurrence risk. This finding needs to be assessed on a larger cohort.
EP
If association between histological IGP and higher local recurrence rates is already well established [7, 31], few studies have investigated links between histological IGP, imaging of STS periphery and distant relapses [7], and none in the setting of NAC. The coexistence of
CC
Oedema, contrast-enhanced oedema, diffuse IGP on MRI may reflect a particular propensity of the tumour to metastasize.
A
The ability to anticipate poor responses during NAC course may help to adjust treatment. Likewise, MRI findings suspecting the presence of satellite tumorous cells in tumour periphery at last evaluation could help surgeons and radiotherapists to adjust their procedure (larger surgical margins, larger radiation fields, systemic adjuvant treatments) or advocate for closer monitoring at the end of treatments. Inter-observer agreements were moderate for most qualitative MRI features although consensus was easily reached between the two experts radiologists. This indicates that visual
11 assessment of surrounding tissue is complex and partly subjective. The retrospective nature of the study has probably biased the reproducibility because oedema was assessed on different MRI sequences, proton density -WI and T2-WI, with different fat suppression techniques which may biased the sensitivity to detect oedema [32, 33]. Similarly, contrast uptake of the surrounding oedema was estimated either on 3D gradient recalled echo or on 2D turbo spin echo fat-sat T1-WI, which may have different contrast-to-noise and signal-tonoise ratio, especially at 1.5 Tesla [34, 35]. It should be noted that focal or diffuse IGP on
IP T
MRI, contrast-enhanced oedema and tail sign were frequently found concomitantly in
patients. Even if we suggested strict definitions of each MRI feature, the distinction was
SC R
sometimes difficult to assess. This may also contribute to moderate reproducibility.
Consequently, further studies should investigate simplification of abnormal peripheryrelated MRI features.
MRI features in this study were all qualitative. Further studies should attempt to introduce
U
quantitative assessment of the surrounding oedema. In our experiment, basic measurements,
N
such as longest diameter of the oedema, were difficult to determine and poorly reproducible
A
because of the complex 3D nature of oedema. Next studies should propose homogeneous and standardized MRI protocols with large Field-of-view T2-WI and T1-WI after
M
Gadolinium-chelates injection with the same fat suppression method, same MRI contrast agent and same acquisition plan throughout the treatment.
TE D
Moreover, other imaging modalities may bring information regarding peritumoral tissue, such as DCE-MRI, DWI and 18FDG/PET-CT and should be added in further studies. Indeed, peritumoral tissues may harbour abnormal growth factors in addition to satellite tumorous
EP
cells resulting in slight quantitative differences in markers derived from their post-treatment. Interestingly, White et al. have demonstrated that satellite single or clusters of tumorous cells could be found beyond the tumour margins without correlation with the size of the
CC
tumour, oedema extension or contrast-enhanced oedema, in population of patients who were only treated by surgery. They found rare cases of satellite tumorous cells without associated
A
oedema [3]. One might hypothesize that more advanced imaging techniques could also identify these satellite tumorous cells without anomalies on conventional MRI, as well as being able to distinguish peritumoral enhancement due to inflammatory infiltrate from satellite tumorous cells. Our study has other limits. Only a limited numbers of patients were included and even a smaller number was used in the survival analysis. We decided to combine metastatic and distant relapses for the disease free survival because patient number and events would have
12 been too small otherwise for statistical analyses. We also simplified the histotypes for adjustment in the multivariate analyses into 2 categories: undifferentiated sarcomas vs. others. Undifferentiated sarcomas, which included undifferentiated pleomorphic sarcomas and myxofibrosarcomas, represent a category of tumours with complex genomic, poor prognosis and frequent abnormal MRI findings in their periphery [10-12]. Dichotomizing histotypes enabled to divide the population in 2 equivalent groups (n=27 vs. n=30) with a pathophysiological meaning. However, larger studies should follow the WHO classification
IP T
in order to better estimate the odds ratio (and hazard ratio) of the MRI features associated
with histological response and satellite tumors cells. To note, other studies that focused on
SC R
imaging biomarkers for STS did not exceed this population size, likely due to the rarity of these tumours [17, 21-24, 36].
Retrospective aspects of the study have also implications on the pathological analysis of the surgical specimen since STS periphery may have been under-sampled. Presence of satellite
U
tumorous cells may have been underestimated. However, we believe that variability in
N
treatments, surgical technics, pathological analyses were limited since patients were
A
managed by experts from a single sarcoma reference centre. Further studies should propose a homogeneous protocol for pathological management of surgical specimen. In that sense,
M
EORTC proposed guidelines for sarcoma pathologists [37] and similar initiatives should focus on tumour periphery. Other known or suspected predictors for OS and DFS were not
TE D
mentioned in the present study, such as C-reactive protein, vascular invasion, 18FDG/PETCT response or modified Choi criteria, and further studies should investigate multivariate models taking into account these co-variables.
EP
To conclude, this study emphasizes the importance of MRI analysis of STS tissue periphery as it may inform about histological response and remaining satellite tumorous cells distant from tumour borders. Further studies are needed to confirm these results and investigate
CC
prospectively STS periphery with standardized multimodal imaging and pathological
A
protocols.
Acknowledgements: The authors would like to thank Mrs Camille Martinerie for medical writing service.
Funding information: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
13
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Conflicts of interest: The authors declare no conflict of interest
14 REFERENCES [1] Honoré C, Méeus P, Stoeckle E, Bonvalot S, Soft tissue sarcoma in France in 2015: Epidemiology, classification and organization of clinical care, J Visc Surg. 152 (2015) 223– 230. [2] Zhao F, Ahlawat S, Farahani SJ, et al., Can MR imaging be used to predict tumor grade in soft-tissue sarcoma? Radiology. 272 (2014) 192–201.
IP T
[3] White LM, Wunder JS, Bell RS, et al., Histologic assessment of peritumoral edema in soft tissue sarcoma, Int J Radiat Oncol Biol Phys. 61 (2005) 1439–1445. [4] Panicek DM, Schwartz LH, Soft Tissue Edema Around Musculoskeletal Sarcomas at Magnetic Resonance Imaging, Sarcoma. 1 (1997) 189–191.
SC R
[5] Carneiro A, Bendahl P-O, Engellau J, et al., A prognostic model for soft tissue sarcoma of the extremities and trunk wall based on size, vascular invasion, necrosis, and growth pattern, Cancer. 117 (2011) 1279–1287.
U
[6] Sampo M, Tarkkanen M, Tukiainen E, et al., A web-based prognostic tool for extremity and trunk wall soft tissue sarcomas and its external validation, Br J Cancer. 106 (2012) 1076–1082.
A
N
[7] Fernebro J, Wiklund M, Jonsson K, et al., Focus on the Tumour Periphery in MRI Evaluation of Soft Tissue Sarcoma: Infiltrative Growth Signifies Poor Prognosis, Sarcoma. 2006 (2006) 21251.
TE D
M
[8] Park JW, Kim H-S, Lee C, Yoo HJ, Yun JY, Han I, Preoperative Factors Associated with Infiltrative Histologic Growth Patterns in Extremity Soft Tissue Sarcoma, Sarcoma. 2017 (2017) 5419394. [9] Nakamura T, Matsumine A, Matsubara T, et al., Infiltrative tumor growth patterns on magnetic resonance imaging associated with systemic inflammation and oncological outcome in patients with high-grade soft-tissue sarcoma, PLoS ONE. 12 (2017).
EP
[10] Kikuta K, Kubota D, Yoshida A, et al., An analysis of factors related to the tail-like pattern of myxofibrosarcoma seen on MRI, Skeletal Radiol. 44 (2015) 55–62.
CC
[11] Lefkowitz RA, Landa J, Hwang S, et al., Myxofibrosarcoma: prevalence and diagnostic value of the “tail sign” on magnetic resonance imaging, Skeletal Radiol. 42 (2013) 809–818.
A
[12] Yoo HJ, Hong SH, Kang Y, et al., MR imaging of myxofibrosarcoma and undifferentiated sarcoma with emphasis on tail sign; diagnostic and prognostic value, Eur Radiol. 24 (2014) 1749–1757. [13] Issels RD, Lindner LH, Verweij J, et al., Neo-adjuvant chemotherapy alone or with regional hyperthermia for localised high-risk soft-tissue sarcoma: a randomised phase 3 multicentre study, Lancet Oncol. 11 (2010) 561–570. [14] Saponara M, Stacchiotti S, Casali PG, Gronchi A, (Neo)adjuvant treatment in localised soft tissue sarcoma: The unsolved affair, Eur J Cancer. 70 (2017) 1–11.
15 [15] Pasquali S, Gronchi A, Neoadjuvant chemotherapy in soft tissue sarcomas: latest evidence and clinical implications, Ther Adv Med Oncol. 9 (2017) 415–429. [16] Gronchi A, Ferrari S, Quagliuolo V, et al., Histotype-tailored neoadjuvant chemotherapy versus standard chemotherapy in patients with high-risk soft-tissue sarcomas (ISG-STS 1001): an international, open-label, randomised, controlled, phase 3, multicentre trial, Lancet Oncol. 18 (2017) 812–822.
IP T
[17] Stacchiotti S, Collini P, Messina A, et al.. High-grade soft-tissue sarcomas: tumor response assessment--pilot study to assess the correlation between radiologic and pathologic response by using RECIST and Choi criteria, Radiology. 251 (2009) 447–456.
SC R
[18] Stacchiotti S, Verderio P, Messina A, et al., Tumor response assessment by modified Choi criteria in localized high-risk soft tissue sarcoma treated with chemotherapy, Cancer. 118 (2012) 5857–5866.
U
[19] Benz MR, Czernin J, Allen-Auerbach MS, et al., FDG-PET/CT imaging predicts histopathologic treatment responses after the initial cycle of neoadjuvant chemotherapy in high-grade soft-tissue sarcomas, Clin Cancer Res Off J Am Assoc Cancer Res. 15 (2009) 2856–2863.
A
N
[20] Benz MR, Allen-Auerbach MS, Eilber FC, et al., Combined assessment of metabolic and volumetric changes for assessment of tumor response in patients with soft-tissue sarcomas, J Nucl Med Off Publ Soc Nucl Med. 49 (2008) 1579–1584.
M
[21] Huang W, Beckett BR, Tudorica A, et al., Evaluation of Soft Tissue Sarcoma Response to Preoperative Chemoradiotherapy Using Dynamic Contrast-Enhanced Magnetic Resonance Imaging, Tomogr J Imaging Res. 2 (2016) 308–316.
TE D
[22] Xia W, Yan Z, Gao X, Volume fractions of DCE-MRI parameter as early predictor of histologic response in soft tissue sarcoma: A feasibility study, Eur J Radiol. 95 (2017) 228– 235.
EP
[23] Soldatos T, Ahlawat S, Montgomery E, Chalian M, Jacobs MA, Fayad LM, Multiparametric assessment of treatment response in high-grade soft-tissue sarcomas with anatomic and functional MR imaging sequences, Radiology. 278 (2015) 831–840.
CC
[24] Meyer JM, Perlewitz KS, Hayden JB, et al., Phase I trial of preoperative chemoradiation plus sorafenib for high-risk extremity soft tissue sarcomas with dynamic contrast-enhanced MRI correlates, Clin Cancer Res Off J Am Assoc Cancer Res. 19 (2013) 6902–6911.
A
[25] Hanna SL, Fletcher BD, Parham DM, Bugg MF, Muscle edema in musculoskeletal tumors: MR imaging characteristics and clinical significance, J Magn Reson Imaging JMRI. 1 (1991) 441–449. [26] Bahig H, Roberge D, Bosch W, et al., Agreement among RTOG sarcoma radiation oncologists in contouring suspicious peritumoral edema for preoperative radiation therapy of soft tissue sarcoma of the extremity, Int J Radiat Oncol Biol Phys. 86 (2013) 298–303.
16 [27] Trojani M, Contesso G, Coindre JM, et al., Soft-tissue sarcomas of adults; study of pathological prognostic variables and definition of a histopathological grading system, Int J Cancer. 33 (1984) 37–42. [28] Eisenhauer EA, Therasse P, Bogaerts J, et al., New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1), Eur J Cancer. 45 (2009) 228–247.
IP T
[29] Cousin S, Crombé A, Italiano A, et al., Clinical, radiological and genetic features associated with the histopathologic response to noadjuvant chemotherapy (NAC) and outcomes in locally advanced soft tissue sarcoma (STS) patients, Journal of Clinical Oncology. 35:15_suppl (2017) 11014-11014. [30] Landis JR, Koch GG, The measurement of observer agreement for categorical data. Biometrics. 33 (1977) 159–174.
SC R
[31] Manoso MW, Pratt J, Healey JH, Boland PJ, Athanasian EA, Infiltrative MRI pattern and incomplete initial surgery compromise local control of myxofibrosarcoma, Clin Orthop. 450 (2006) 89–94.
N
U
[32] Delfaut EM, Beltran J, Johnson G, Rousseau J, Marchandise X, Cotten A, Fat suppression in MR imaging: techniques and pitfalls, Radiogr Rev Publ Radiol Soc N Am Inc. 19 (1999) 373–382.
A
[33] Bley TA, Wieben O, François CJ, Brittain JH, Reeder SB, Fat and water magnetic resonance imaging, J Magn Reson Imaging JMRI. 31 (2010) 4–18.
M
[34] Hargreaves BA, Rapid gradient-echo imaging, J Magn Reson Imaging JMRI. 36 (2012) 1300–1313.
TE D
[35] Zur Y, Wood ML, Neuringer LJ, Spoiling of transverse magnetization in steady-state sequences, Magn Reson Med. 21 (1991) 251–263.
EP
[36] Favinger JL, Hippe DS, Davidson DJ, et al., Soft Tissue Sarcoma Response to Two Cycles of Neoadjuvant Chemotherapy: A Multireader Analysis of MRI Findings and Agreement with RECIST Criteria and Change in SUVmax, Acad Radiol. 25 (2018) 470– 475.
A
CC
[37] Wardelmann E, Haas RL, Bovée JVMG, et al., Evaluation of response after neoadjuvant treatment in soft tissue sarcomas; the European Organization for Research and Treatment of Cancer–Soft Tissue and Bone Sarcoma Group (EORTC–STBSG) recommendations for pathological examination and reporting, Eur J Cancer. 53 (2016) 84– 95.
17 LEGENDS
SC R
IP T
Figure 1. MRI features depicting surrounding tissue of soft-tissue sarcoma. (A) ‘Pushing-type’ growth pattern (: entirely well-defined tumour) in a grade III undifferentiated round cells sarcoma of the thigh. (B) ‘Focal-type’ infiltrative growth pattern (: irregular borders and infiltration of surrounding tissue <25% of tumour circumference, white arrow) in a grade III pleomorphic liposarcoma of the thigh. (C) ‘Diffuse-type’ infiltrative growth pattern (: irregular borders and infiltration of surrounding tissue ≥25% of tumour circumference) in a grade III undifferentiated epitheloid sarcoma of the arm. (D) Aponeurotic enhancement of ≥2mm thickness (‘tail sign’, white arrows) in a grade III myxofibrosarcoma of the arm. (E) Surrounding oedema (white arrow) without contrast enhancement in a grade III undifferentiated pleomorphic sarcoma of the psoas muscle. (F) Contrast enhancement of a surrounding oedema (white arrow), without mass effect in a grade III dedifferentiated liposarcomas of the thigh. T2: T2 weighted imaging, FS T1+: Fatsuppressed T1 weighted imaging after Gadolinium-chelates injection, FS T2: T2 weighted imaging with fat suppression.
A
N
U
Figure 2. Kaplan-Meier curves of MRI and pathological features to predict diseasefree survival. Significant separation was determined by log rank test, *: p<0.05, **: p<0.005
A
CC
EP
TE D
M
Figure 3. Radio-pathological correlation of a good histological responder with persistent satellite tumorous cells after neoadjuvant chemotherapy. Patient presented with a deeply-located grade III leiomyosarcoma of the left quadriceps femori. (A) MRI at baseline and (B) early evaluation: during neoadjuvant chemotherapy, tumor demonstrated strong decrease of T2 signal intensity compatible with a fibrotic response (white arrow), together with foci of necrosis. However, surrounding oedema rather tended to increase (white arrow-head) and still displayed marked enhancement after Gadolinium-chelates injection (lined white arrow-head). (C) Pathological analysis of the surgical specimen confirmed a good response with 5% stainable cells, 85% fibrosis (Δ), 10% necrosis and demonstrated spans of fibrosis (°) between muscular bundles (§) but magnification (black frame) showed single and clusters of satellite tumorous cells beyond the tumour borders (black arrow). Follow-up demonstrated lung metastatic relapse 42 months after surgery and patient died of disease 4 months later. FS PD: Fat-Sat proton density weighted imaging, T2: T2 weighted imaging, FS T1+: Fat-Sat T1 weighted-imaging after Gadolinium-chelates injection. Figure 4. Example of a good responder without satellite tumour cells on surgical specimen: classical evolution of the surrounding tissue during neoadjuvant chemotherapy. Patient presented with a deep and superficial grade III undifferentiated pleomorphic sarcoma of the thigh. (A) Baseline MRI showed a diffuse infiltrative growth pattern with poorly defined margin on >25% tumour circumference, associated with surrounding oedema (white arrow) that enhanced after Gadolinium-chelates injection, and
18 with thick aponeurotic enhancement spreading along the superficial fascia (white arrow head). After 2 cycles of chemotherapy (B), these peripheral anomalies decreased in an unequivocal manner and >50% of tumour volume demonstrated fibro-necrotic changes (white asterix). (C) Pathological analysis confirmed a good response. Nine years later, patient is alive in remission. T2: T2-weighted imaging, FS T1+: Fat-Sat T1-weighted imaging after Gadolinium-chelates injection.
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Figure 5. Example of a good responder without satellite tumorous cells on surgical specimen: atypical evolution of the surrounding tissue during neoadjuvant chemotherapy. Patient presented with a superficial and aponeurotic grade III myxofibrosarcoma of the upper limb. (A) Tumour initially showed a focal infiltrative growth pattern. At early evaluation (B) and late evaluation (C), oedema tended to increase (white arrow), with additional enhancement in the subcutaneous fat and increased aponeurotic enhancement (white arrowhead). However, tumour demonstrated extensive fibro-necrotic changes suggestive of a good response with occurrence of fibrotic capsule (dashed white arrow). These findings were still observed at late evaluation. (D) Pathological analysis confirmed a good histologic response and found a thick fibrotic capsule (°) with two fronts of inflammatory cells (*) made of lymphocytes, histiocytes, inside and outside, as well as activated fibroblasts on magnification (black frame). No satellite tumorous cell was found and we concluded that the increased surrounding anomalies were due to an inflammatory reaction.
EP
CC
A TE D
IP T
SC R
U
N
A
M
19
EP
CC
A TE D
IP T
SC R
U
N
A
M
20
EP
CC
A TE D
IP T
SC R
U
N
A
M
21
EP
CC
A TE D
IP T
SC R
U
N
A
M
22
23 Table 1. Population study.
21 (36.8) 36 (63.2)
Age (in years)
58 (31-77)
Histotypes Undifferentiated pleomorphic sarcoma Myxofibrosarcoma Rhabdomyosarcoma Leiomyosarcoma M/RC-LPS Dedifferentiated liposarcomas Pleomorphic liposarcoma Synovial sarcoma MPNST
22 (38.6) 5 (8.8) 8 (14) 5 (8.8) 6 (10.5) 1 (1.8) 3 (5.3) 6 (10.5) 1 (1.8)
M
A
2 (3.5) 41 (71.9) 14 (24.6)
TE D
Situation Shoulder Girdle Upper Limb Trunk Wall Pelvic Girdle Lower Limb
N
Depth Superficial Deep Deep + Superficial
LD at baseline (in mm)
EP
U
Gender Female Male
IP T
No of patients
SC R
Characteristics
Time last cycle to Surgery (in days)
6 (10.5) 6 (10.5) 8 (14) 2 (3.5) 35 (61.4) 102 (39-260) 31 (17-64)
A
CC
NOTE - LD : longest diameter. M/RC-LPS : Myxoid / round cells liposarcoma, MPNST : malignant peripheral nerve sheath tumor. Results are given as numbers of patients with percentage in parentheses, except for age, longest diameter at baseline and time from last cycle to surgery, which are given as median and range in parentheses.
24 Table 2. Inter-observer agreement of the MRI features and their changes from baseline (MRI-0) to early (MRI-1) and pre-surgery evaluation (MRI-2). Table 2. Inter-observer agreement of the MRI features and their changes from baseline MRI features
κ = 0.597, (0.217-0.877)
p=0.000
Contrast-enhanced oedema
κ = 0.624, (0.391-0.857)
p=0.000
Aponeurotic enhancement
κ = 0.602, (0.404-0.800)
p=0.000
Bone, vessel or nerve invasion
κ = 0.855, (0.720-0.990)
p=0.000
wκ = 0.427, (0.239-0.615)
p=0.000
Change in margin definition
κ = 0.335, (0.119-0.551)
p=0.003
Change in oedema
κ = 0.527, (0.305-0.749)
p=0.000
Change in contrast-enhanced oedema
κ = 0.553, (0.349-0.757)
p=0.000
Change in Aponeurotic enhancement
κ = 0.304, (0.049-0.559)
p=0.021
Change in margin definition
κ = 0.243, (0.027-0.459)
p=0.049
Change in oedema
κ = 0.634, (0.372-0.897)
Change in aponeurotic enhancement
A
Change in contrast-enhanced oedema
p=0.000
κ = 0.299, (0.001-0.597)
p=0.042
κ = 0.211, (0-0.556)
p=0.183
N
MRI-2 (N=36)
SC R
Growth pattern MRI-1 (N=57)
IP T
Oedema
U
MRI-0 (N=57)
Inter-Observer Agreement (CI95%, p-value)
M
(MRI-0) to early (MRI-1) and pre-surgery evaluation (MRI-2).
A
CC
EP
TE D
NOTE – Abbreviations : CI95% : 95% confidence interval, κ : Cohen’s kappa ; LD : longest diameter, wκ : weighted kappa.
25
Table 3. Associations between pathological findings on surgical specimen and MRI features at baseline (MRI-0), early (MRI-1) and late evaluations (MRI-2). Histological Response Good
Poor
OR (CI95%)
Satellite tumor cells p-value
No
Yes
OR (CI95%)
p-value
MRI-0 Oedema
No
1
5
_
Yes
14
37
0.53 (0.06-4.93)
No
4
12
_
3
3
_
25
22
0.88 (0.16-4.82)
10
5
_
18
20
2.22 (0.64-7.74)
15
10
11
15
16
18
12
7
8
1
_
10
7
9.60 (1.10-83.40) 1
10
17
3.82 (1.122-11.98) 2
0
0
_
1
2
_
24
21
_
1.000
Aponeurotic enhancement
1.000 Yes
11
30
0.91 (0.24-3.42)
No
7
19
_
0.237
1.000
MRI growth pattern
7
22
1.16 (0.34-3.90)
No
10
25
_
Yes
5
17
1.36 (0.39-4.69)
Pushing
1
8
_
0.761
1
Focal infiltration
4
14
0.30 (0.03-2.66)
Diffuse
10
20
0.46 (0.13-1.56) 2
CR
0
0
_
PR
1
1
_
Oedema 3
Contrast enhanced oedema Aponeurotic enhancement 3
PD
1
4
Better delimited
4
8
Stable or worst
11
Decrease or absent
9
Increase or stable
6
Decrease or absent
11
Contrast-enhanced oedema Aponeurotic enhancement
11
_
30
4.09 (1.18-14.18)
15
_ 4.95 (1.34-18.29)
Decrease or absent
11
23
_
_
0.030 *
0.017 *
3
2
_
11
1
_
17
24
15.53 (1.83-131.9)
13
6
15
18
2.60 (0.79-8.51)
18
6
_
10
19
5.70 (1.72-18.92)
20
12
_
19
3.03 (0.74-12.45)
7
13
3.09 (0.97-9.92)
CR
0
0
_
0
0
_
3
3
_
3
2
_
0.153
0.326
SD
6
22
_
14
12
_
PD
0
3
_
0
2
_
Better delimited
5
7
10
2
_
7
14
10.00 (1.71-56.63)
21
3.75 (0.78-17.99)
Decrease or absent
9
13
_
Increase or stable
0
15
0 (0-0.47)
Decrease or absent
8
15
_
0.005 **
13
6
_
4
10
5.42 (1.20-24.52)
13
7
_
4
9
4.18 (0.94-18.16)
15
9
_
2
7
5.833 (0.988-34.436)
0.112 Increase or stable
1
13
6.93 (0.76-63.05)
Decrease or absent
7
20
_
2
8
1.40 (0.24-8.24)
0.010 *
0.037 *
0.080
1.000 Increase or stable
0.005 **
0.087
3
4
0.003 **
0.152
Increase or stable
Stable or worst
0.022 *
0.754
0.205
0.116
0.390
0.52 (0.16-1.64)
0.713
1.55 (0.39-6.14)
27
EP CC
A
Oedema
_
4
PR
Margin definition
_
0.267
2.05 (0.670-6.24)
0.710
_
Increase or stable
MRI-0 to MRI-2 RECIST 1.1
34
N
37
A
13
M
Margin definition
SD
TE D
RECIST 1.1
0.370
U
MRI-0 to MRI-1
_
SC R
Bone, vessel or nerve invasion
Yes
IP T
Contrast-enhanced oedema
1.000
0.057
NOTE – Abbreviations: CI95%: 95% confidence interval, CR : complete response, OR: odds ratio, PD : progressive disease, PR : partial response, SD : stable disease. MR features with 3 values were dummying as follows : 1 : Pushing growth pattern versus presence of infiltration, 2 : pushing or focal infiltration versus infiltrating. 3 : At early evaluation, one patient was removed for the assessment of
26 change in contrast-enhanced oedema and one another patient for aponeurotic enhancement as STIR T2-WI and post-Gd FS T1-WI did not capture the whole abnormal findings, thus the statistical analysis was made on 56 patients instead of 57. * : p<0.05, **, p<0.005.
Table 4. Multivariate analysis of the MRI features associated with pathological findings Histological Response OR (CI95%)
Satellite tumor cells
p-value
OR (CI95%)
p-value
MRI-0 Pushing or Focal
_
Diffuse
_
better delimited
_
_
_
3.52 (1.06-11.67)
MRI-0 to MRI-1
Oedema
Stable or worst
_
Decrease or absent
_
Increase or stable Contrast-enhanced oedema
_ _
_
0.011 *
6.87 (1.56-30.33)
Decrease or absent
_
Increase or stable
8.06 (1.74-37.32)
19.06 (1.97-185.6)
SC R
Margin definition
0.039 *
IP T
MRI growth pattern
_
0.011 *
_
_
0.008 *
5.59 (1.54-20.36)
0.009 *
A
CC
EP
TE D
M
A
N
U
NOTE – Abbreviations: CI95%: 95% confidence interval, OR: odds ratio. Results were adjusted for histotype (myxofibrosarcoma or undifferentiated sarcoma versus others), age (< or ≥ median age = 58 years old) and size at baseline (< or ≥ median size = 102 mm). *: p<0.05.
27 Table 5: Univariate survival analysis of the MRI features evaluated at baseline (MRI-0), at early evaluation (MRI-0 to MRI-1) and late evaluation (MRI-0 to MRI-2) Overall Survival Survival (CI95%)
HR (CI95%)
No Yes No Yes No Yes No Yes Pushing or focal Diffuse
94.6 (51.1-138.2) 70.3 (53-87.6) 51 (33.8-28.3) 82.1 (61-103.2) 66.9 (36.1-97.8) 48.7 (34-63.5) 77 (50.7-103.2) 47.4 (34.8-60) 94.9 (69.3-120.6) 45.8 (32.6-59)
_ 1.51 (0.19-12.13) _ 0.89 (0.25-3.25) _ 1.22 (0.41-3.66) _ 1.16 (0.38-3.54) _ 2.18 (0.67-7.11)
Objective response No response Better delimited Stable or worst Decrease or absent Increase or stable Decrease or absent Increase or stable Decrease or absent Increase or stable
34.5 (19.6-49.5) 76.9 (56-97.8) _£ 67.1 (44.7-89.5) 79.7 (53-106.4) 66.1 (41.4-90.9) 82.4 (57.3-107.6) 63.8 (38.4-89.3) 83.7 (53.7-114) 42.3 (24.6-60)
_ 0.73 (0.09-5.72) _ 28.99 (0.07-12437) _ 2.88 (0.79-10.51) _ 3.46 (0.94-12.68) _ 2.92 (0.94-9.01)
Objective response No response Better delimited Stable or worst Decrease or absent Increase or stable Decrease or absent Increase or stable Decrease or absent
34.5 (19.6-49.5) 39.5 (29.8-49.3) 86.6 (46.1-127.1) 38.9 (29.1-48.8) 93.3 (72.3-81) 25.5 (15-35.9) 85.7 (61.8-109.5) 30.9 (17-44.8) 84.5 (62.9-106)
_ 0.99 (0.12-8.09) _ 2.09 (0.26-17.15) _ 5.69 (1.13-28.77) _ 3.45 (0.81-14.73) _
Increase or stable
19.5 (11.4-127.6)
5.77 (1.24-26.91)
Disease-free Survival p-value
Survival (CI95%)
HR (CI95%)
84.3 (26.5-142) 48.1 (30.5-65.8) 32.1 (16.3-47.9) 57.8 (35.7-97.9) 56.3 (28.2-84.3) 37.6 (14-41.3) 68.9 (36.7-90.9) 30.5 (16.6-44.3) 83.2 (53.6-113) 23.3 (12.2-33.8)
_ 2.02 (0.27-15.24) _ 0.96 (0.32-2.89) _ 1.55 (0.62-3.89) _ 1.47 (0.59-3.63) _ 3.72 (1.31-10.54)
p-value
Margin definition Oedema Contrast-enhanced oedema Aponeurotic enhancement
MRI-0 to MRI-2 RECIST1.1 Margin definition Oedema Contrast-enhanced oedema Aponeurotic enhancement
0.787 0.184
SC R
MRI-0 to MRI-1 RECIST1.1
0.721
0.763
0.080 0.095
U
Growth pattern
TE D
bone, vessel or nerve invasion
0.861
0.052
N
Aponeurotic enhancement
0.697
0.052
A
Contrast-enhanced oedema
M
Oedema
IP T
MRI-0
0.993 0.481 0.019 * 0.077 0.012 *
19.3 (0-43.6) 57.9 (37.2-78.6) _£ 43.7 (24.2-63.2) 40.4 (43.8-97) 35.8 (13.6-58) 60.4 (36.1-84.8) 47.7 (22-73.4) 65.9 (93.7-92.1) 26.9 (9.2-44.6)
_ 0.45 (0.10-1.99) _ 30.74 (0.25-3788) _ 2.80 (1.01-7.84) _ 1.85 (0.72-4.73) _ 2.37 (0.95-5.89)
19.3 (0-43.6) 28.6 (18-39.3) 85 (41.8-128) 24.7 (14.8-34.6) 67.1 (40.2-94) 14.2 (5-23.5) 59.2 (31.8-86.6) 1.7 (5.7-29.3) 57.6 (33.5-81.7)
_ 0.52 (0.11-2.43) _ 5.85 (0.69-43.42) _ 4.98 (1.42-17.40) _ 2.93 (0.94-9.09) _
9.2 (1.8-16.6)
6.06 (1.57-23.44)
A
CC
EP
NOTE – Abbreviations: CI95%: 95% confidence intervals, HR: hazard ratio, LD: longest diameter. Mean survivals are given in months. HR and survivals are given with CI95%. *: p<0.05, **: p<0.005, £: not calculable as only 4 patients displayed this pattern and none of them died or had an event related to disease.
0.486 0.938 0.340 0.404 0.009 *
0.279 0.08 0.040 * 0.191 0.056
0.398 0.073 0.006 * 0.054 0.003 **
S0720-048X(18)30399-1 https://doi.org/10.1016/j.ejrad.2018.11.004 EURR 8364
To appear in:
European Journal of Radiology
Received date: Revised date: Accepted date:
24 June 2018 3 October 2018 4 November 2018
Please cite this article as: Crombe A, Le Loarer F, Stoeckle E, Cousin S, Michot A, Italiano A, Buy X, Kind M, MRI Assessment of Surrounding Tissues in Soft-tissue Sarcoma During Neoadjuvant Chemotherapy Can Help Predicting Response and Prognosis, European Journal of Radiology (2018), https://doi.org/10.1016/j.ejrad.2018.11.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
MRI Assessment of Surrounding Tissues in Soft-tissue Sarcoma During Neoadjuvant Chemotherapy Can Help Predicting Response and Prognosis Amandine CROMBE1,2,3, François LE LOARER3,4, Eberhard STOECKLE5, Sophie COUSIN6, Audrey MICHOT5, Antoine ITALIANO6, Xavier BUY1, Michèle KIND1.
IP T
1. Department of Radiology, Institut Bergonie, F-33000 Bordeaux, France 2. INRIA Bordeaux-Sud-Ouest, CNRS UMR 5251, F-33405 Talence, France 3. Université de Bordeaux, F-33000, Bordeaux, France
SC R
4. Department of Pathology, Institut Bergonie, F-33000 Bordeaux, France 5. Department of Surgery, Institut Bergonie, F-33000 Bordeaux, France
U
6. Department of Medical Oncology, Institut Bergonie, F-33000 Bordeaux, France
Corresponding author: Dr Amandine Crombé
N
email: [email protected]
A
fax: +33 (0) 5 56 33 33 30
M
tel: +33 (0) 5 56 33 33 33
address: Department of Radiology, Institut Bergonié
HIGHLIGHTS
Changes in surrounding tissues of STS during chemotherapy are associated with
EP
TE D
229 cours de l’Argonne, 33000 Bordeaux, France
response.
Infiltrative growth pattern and stable/increased oedema may predict worse outcome.
CC
Evaluation of surrounding tissues may help anticipating persistence of satellite
A
tumorous cells
ABSTRACT Objectives: To determine if changes in surrounding tissues of soft-tissue sarcomas (STS) evaluated by MRI during neoadjuvant chemotherapy (NAC) are associated with the histological response and satellite tumorous cells beyond the pseudocapsule on surgical specimen, disease-free survival (DFS) and overall survival (OS).
2 Methods: Fifty-seven adult patients with locally advanced high-grade STS of extremities and trunk wall were included in this single-centre retrospective study. All were uniformly treated by 5-6 cycles of anthracycline-based NAC, curative surgery and adjuvant radiotherapy and had available MRI with a gadolinium-chelates administration at baseline and after 2 cycles. Thirty-seven patients also had a pre-operative MRI. Two senior radiologists evaluated MRI growth pattern, oedema, contrast-enhanced oedema, aponeurotic enhancement, and their qualitative changes during NAC. An expert pathologist reviewed all
IP T
surgical specimens. A good histological response was defined as <10% viable cells.
Associations with pathological findings were estimated with Fisher and Chi-square tests and
SC R
multivariate analysis with binary logistic regression. Survival analyses included log-rank tests.
Results: Forty-two patients had poor responses and 25 had satellite tumorous cells on surgical specimen. Changes in surrounding oedema and in contrast-enhanced oedema were
U
associated with responses (p=0.008 and 0.011, respectively). Diffuse infiltrative growth
N
pattern (IGP) on baseline MRI, changes in margin definition and in contrast-enhanced
A
oedema at early evaluation were associated with satellite tumorous cells (p=0.039, 0.011 and 0.009, respectively). Diffuse IGP on baseline MRI and stable or increased oedema during
M
treatment were predictors of DFS (p=0.009 and 0.040, respectively) Conclusion: Surrounding tissues of STS during NAC should be carefully evaluated as they
TE D
may steer treatment efficacy and patient prognosis.
ABBREVIATIONS
EP
CI95%: 95% confidence interval
DCE-MRI: dynamic contrast enhanced MRI DFS: disease free survival
CC
DWI: diffusion weighted imaging FNCLCC: French ‘Federation Nationale des Centres de Lutte Contre le Cancer’
A
HR: Hazard ratio IGP: infiltrative growth pattern LD: longest diameter MPNST: malignant peripheral nerve sheath tumour M/RC-LPS: myxoid/round cells liposarcoma NAC: neoadjuvant chemotherapy OR: Odds ratio
3 OS: overall survival STS: soft tissue sarcoma WI: weighted imaging
KEYWORDS Soft-tissue sarcoma ; Peritumoral tissue ;
IP T
Infiltrative growth pattern ; Chemotherapy ;
SC R
Tumour response evaluation;
1. INTRODUCTION
Soft-tissue sarcomas (STS) represent a heterogeneous group of ubiquitous mesenchymal
U
tumours. About 4000 new cases of STS are annually diagnosed in France, making these
N
tumours the most frequent of rare tumours [1]. On MRI, their periphery can demonstrate a
A
wide range of anomalies. Oedema in surrounding tissue is frequently observed and might be related to high-grade tumours and to isolated or clusters of satellite tumorous cells beyond
M
the tumour borders [2-4]. Diffuse infiltrative growth pattern (IGP) on MRI, which consists in poorly defined margins, has also been associated with high-grade STS and with
TE D
histological IGP - an independent predictor of poorer prognosis [5-8]. Nakamura et al. also recently highlighted that diffuse IGP on MRI was predictive of disease–free survival (DFS) and metastasis-free survival [9]. Additionally, the ‘tail sign’, which corresponds to a thick
EP
aponeurotic enhancement after Gadolinium-chelates injection due to tumour spreading along fascia in myxofibrosarcomas and undifferentiated pleomorphic sarcoma, should be carefully removed during surgery in order to decrease the risk of local relapse [10-12].
CC
Standard of care for high-grade STS is on the verge of being redefined since anthracyclinebased neoadjuvant chemotherapy (NAC) has shown promising results on outcome with
A
more conservative surgical procedures [13-16]. Imaging evaluation of responses to treatment remains based on RECIST 1.1 despite its limited ability to predict histological response [17]. Imaging markers, such as modified Choi criteria, DCE-MRI, 18FDG/PET-CT scan, are superior to RECIST 1.1 but none has so far been used to assess tumour periphery and their therapy-related changes [17-24]. A single sub-study of 6 cases has previously linked surrounding oedema to poorer responses to chemotherapy [25]. Moreover, systematic radio-
4 pathological correlations within tumour periphery might help to adjust the surgical margins and the radiation fields for post-operative radiotherapy [26]. Hence, we aimed at (i) characterizing STS periphery with MRI at baseline and changes during NAC, (ii) correlating the MRI findings with histological response and persistence of viable satellite tumorous cells on the surgical specimen and survival, and (iii) identifying links between these MRI findings and patient outcome.
IP T
2. METHODS 2.1. Population and design of the study
SC R
Our institutional review board approved this retrospective single centre study and informed consent was waived. Between January 2008 and June 2017, we included consecutive patients treated at our institution as they presented with newly diagnosed biopsy-proven high-grade STS of trunk wall or extremities, without metastasis (assessed on whole body
U
MRI for myxoid/round cells liposarcoma and chest CT-scan for other histotypes), uniformly
N
treated with 5-6 cycles of anthracycline-based NAC followed by a curative surgery and an
A
adjuvant radiotherapy, with available baseline MRI (MRI-0) and MRI after 2-3 cycles (MRI-1). Grade was defined according to the French Fédération Nationale des Centres de
M
Lutte Contre le Cancer (FNCLCC) grading system and high-grade corresponded to grade III tumours [27].
TE D
Following co-variables were reported: age, gender, histotype, tumour location, depth relative to superficial fascia, delay from last cycle to surgery, surgical margins, adjuvant radiotherapy, occurrence of relapse (local or metastatic) or tumour-related death. At our
EP
institution, routine follow-up consists in clinical examination and chest radiograph every 3 months for 2 years, every 6 months for 5 years and then annually until 10 years after surgery. These examinations were complemented by chest CT scans (with 1mm thick
CC
reconstruction) and MRI for local evaluation in case of abnormal or doubtful findings. All
A
local or distant relapses were histopathologically confirmed.
2.2. MRI analysis We only included MRI with at least two orthogonal acquisition plans and following available sequences: (i) T2-weighted-imaging (-WI), (ii) fat suppressed T2-WI or proton density sequences and (iii) T1-WI after Gadolinium-chelates injection. All techniques to suppress fat were accepted: short TI inversion recovery, Fat-sat, Dixon methods. MRI were all performed on 1.5T MR system, mostly on the same system (Magnetom AERA 1.5T,
5 Siemens Healthineers, Erlangen, Germany) except for 30 baseline MRI that were performed in external radiological centre before the patient had been addressed to our sarcoma reference centre. Details for conventional MRI protocol performed at our institution are given in Supplementary Data. MRI analysis was performed by 2 senior radiologists from a sarcoma reference centre (with 4 and 30 years of experience in musculoskeletal tumours, respectively), by using a picture archiving and communication system workstation (Entreprise Imaging, AGFA, the
IP T
Netherlands). Each radiologist independently read the whole sets of MRI blinded to
pathological and radiological reports. A final consensual lecture was made between the two
SC R
radiologists in case of disagreement.
All patients had baseline MRI-0 (n=57), MRI-1 (n=57). Additionally, 37 patients had a preoperative MRI, at the end of NAC (called MRI-2).
The following features depicting tumour periphery were reported (Fig. 1):
U
- Growth pattern at baseline on whole tumour circumference on T2-WI and T1-WI after
N
gadolinium-chelates injection and fat suppression. It was divided in 3 categories according to Nakamura et al. [9]: ‘pushing-type’ (i.e. when the tumour was entirely well-defined),
A
‘focal-type’ (i.e. when infiltrative pattern, namely irregular borders and infiltration of
tumour circumference).
M
surrounding tissue represented <25% of tumour circumference) and ‘diffuse-type’ (≥ 25% of
TE D
- Bone, vessel and/or nerves invasion at baseline - Presence of surrounding oedema at baseline and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2 (absent or decrease versus stable or increase).
EP
Oedema was defined as ill-defined non-fatty area of bright SI on T2-WI or proton density WI beyond tumour borders, without mass effect. - Contrast uptake of the surrounding oedema (referred as ‘contrast-enhanced oedema’ in this
CC
study) at baseline and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2. Similarly, qualitative assessment was dichotomised as: absent or decrease versus
A
stable or increase. - Aponeurotic enhancement of more than 2mm thickness at baseline (referred as ‘tail sign’ in the literature [10-12] and qualitative changes during NAC from MRI-0 to MRI-1 and MRI-0 to MRI-2, evaluated on T1-WI after Gadolinium-chelates injection and fat suppression. The radiologists also reported longest diameters at MRI-0, MRI-1 and MRI-2 (tumour-LD, in mm) and relative change from MRI-0 to MRI-1 and from MRI-0 to MRI-2 (in %), in
6 order to assess response status according to RECIST 1.1 [28]. Measures were performed on T2-WI, helped by all the other available MR sequences.
2.3. Pathological analysis Haematoxylin and eosinophil stained (HES) slices of all surgical specimens were retrospectively reviewed for the study by an expert pathologist from our sarcoma reference center, blinded to clinical course. The pathologist reported:
IP T
- The histological response with estimation of the percentage of viable tumour (stainable
cells), necrotic and post-therapy fibrosis on the entire tumour volume. A good response was
SC R
defined with the threshold of <10% viable cells, as it has been demonstrated to be predictive of worse metastatic-free survival and OS [29].
- Presence of satellite tumorous cells beyond tumour borders and pseudocapsule if present. Of note, 4 surgical specimens had not been sampled according to guidelines thus not
U
enabling an assessment of the whole periphery of the tumours. These cases were
N
consequently removed from statistical analyses focusing on the presence of satellite
A
tumorous cells as covariables or outcome. Surgical margins were classified by consensus between surgeon and pathologist. They were classified as R0 (i.e. macroscopically complete
M
with negative microscopic margins), R1 (i.e. macroscopically complete with positive
TE D
microscopic margin) and R2 (i.e. macroscopically incomplete).
2.4. Statistical analysis
Inter-observer agreements were assessed using the Cohen’s Kappa (κ) test for dichotomised variables. Weighted Kappa statistics were used for ordinal variables. The agreement for
EP
classical κ and κw was defined as slight (0-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80) and almost perfect (0.81-0.99) [30].
CC
Associations between categorical (or ordinal variables) and histological findings were performed with the Chi-2 test and Fisher test, as appropriate. A binary multivariate logistic
A
regression modelling was built to identify the MRI features that remained predictive of the histological findings. Only features with a p-value of less than 0.05 were included in the multivariate analysis. Patients with missing values were removed from the multivariate analysis. We added the following dichotomized covariables for adjustment in the multivariate analysis: histotype (undifferentiated sarcoma, which corresponded to undifferentiated pleomorphic sarcomas and myxofibrosarcomas, versus others), age (< or ≥ median age) and size (< or ≥ median size). A multivariate analysis including several
7 radiological parameters altogether was not carried out because of the limited population size. Univariate and multivariate odds ratio (OR) are given with their 95% confidence interval. Survival analysis included the overall survival (OS) and disease-free survival (DFS) in months since the curative surgery was performed. Local recurrence-free survival and metastasis-free survival were combined because of the limited population size and the small number of events. Univariate analyses were made with Kaplan-Meier log-rank tests. The hazard ratio (HR) with CI95% was calculated to measure the degree of survival differences
IP T
on the Kaplan-Meier plots. We excluded patients without any event during a follow-up of less than 2 years after curative surgery.
SC R
All tests were two-tailed. Statistical analyses were done using the SPSS statistical package (IBM, version 21.0, Chicago, IL). Variables are expressed as mean, standard deviation,
U
median and range, as appropriate. A p-value < 0.05 was deemed significant.
N
3. RESULTS
A
3.1. Population (Table 1)
Fifty-seven patients met all the inclusion criteria (21 females, median age 58 years old,
M
range: 31-77) among the 283 patients with newly diagnosed sarcoma who received NAC at our institution during the study period. Twenty-two (38.6%) had an undifferentiated
TE D
pleomorphic sarcoma, 5 (8.8%) had a myxofibrosarcoma, 8 (14%) had a rhabdomyosarcoma, 5 (8.8%) had a leiomyosarcoma, 3 (5.3%) had a pleomorphic liposarcoma, 1 (1.8%) had a dedifferentiated liposarcomas, 6 (10.5%) had myxoid/round cell
EP
liposarcomas, 6 (10.5%) had a synovial sarcoma and one presented with a malignant peripheral nerve sheath tumour (MPNST). The median size before treatment was 102 mm (range: 39–260). Only 2 were superficially situated (3.5%). Twenty patients did not have a
CC
pre-operative MRI for late evaluation.
A
3.2. Inter-observer agreements of the MRI features (Table 2) Inter-observer agreements for MRI features at baseline, early and late evaluations were all statistically significant except for the qualitative change in aponeurotic enhancement between MRI-0 and MRI-2 (κ=0.211, p=0.183). At baseline, all features depicting tumour periphery were substantially reproducible (κ>0.600) except for growth pattern, which was moderately reproducible (κ=0.427). At early evaluation, change in oedema and contrastenhanced oedema were moderately reproducible (κ=0.527 and 0.553, respectively). Change
8 in margin definition was also fairly reproducible at both early and late evaluations (κ=0.335 and 0.243, respectively).
3.3. Association with pathological findings Association with histological response (Table 3) Forty-two (73.7%) patients were poor responders. None of the baseline MRI findings was associated with the response.
IP T
Stability or increase of a pre-existing oedema at early evaluation was seen in 40% (6/15) good responders and 73.2% (30/41) in poor responders. It was associated with a poor
SC R
response at univariate analysis (OR=4.09, p=0.030).
Stability or increase of a pre-existing contrast-enhanced oedema at early evaluation was seen in 26.7% (4/15) good responders and 64.3% (27/42) poor responders, being associated of a poor response at univariate analysis (OR=4.95, p=0.017).
U
At multivariate analysis, both stable or increased oedema and contrast-enhanced oedema at
N
early evaluation remained predictors of a poor response after adjustment for age, size and
A
histotypes (OR=6.87, p=0.011 and OR=8.06, p=0.008, respectively - Table 4) Patients were mostly classified as stable disease according to RECIST 1.1: 86.7% (13/15)
M
good responders and 88.1% (37/42) poor responders at early evaluation, and 66.7% (6/9) good responders and 78.6% (22/28) poor responders at late evaluation. RECIST 1.1 did not
respectively).
TE D
correlate with histological response at early and late evaluation (p=0.710 and 0.153,
Association with satellite tumorous cells beyond tumour border
EP
Twenty-five out of 53 (47.2%) patients had viable satellite tumorous cells beyond tumour borders on surgical specimen. MRI growth pattern at baseline was associated with satellite tumorous cells (p=0.022). Seventeen out of 25 (68%) patients with satellite tumorous cells
CC
and 10 out of 28 (35.7%) patients without had diffuse infiltrative growth pattern on MRI. At early evaluation, stable or worst margin definition was associated with satellite tumorous
A
cells (OR=15.53, p=0.003). This feature was seen in 96% (24/25) of patients with satellite tumorous cells on surgical specimen and 60.7%, (17/28) of patients without. It remained significantly associated at late evaluation (OR=10.00, p=0.010). Additionally, 76% (19/25) of patients with satellite tumorous cells and 35.7% (10/28) of patients without had a stable or increased contrast-enhanced oedema at early evaluation. This MRI feature was associated with satellite tumorous cells at early evaluation (OR=5.70, p=0.005) and also at late evaluation (OR=5.42, p=0.037).
9 At multivariate analysis, diffuse IGP on baseline MRI, stability or worsening of margin definition, and stable or increased contrast-enhanced oedema remained associated with satellite tumorous cells on surgical specimen after adjustment for age, size and histotypes (OR=3.52, p=0.039; OR=19.06, p=0.011, and OR=5.59, p=0.009, respectively – Table 4).
3.4. Prognostic value of radio-pathological findings involving surrounding tissue (Table 5) Survival analysis was carried on 31 patients. The mean survival of the series was 76 months
IP T
(CI95%=(56-97)). Thirteen out of 31 patients (41.9%) had died of the disease and one died of other causes, 19 out of 31 patients (61.3%) had a relapse, either local (7/31, 22.6%) or
SC R
distant (17/31, 54.8%), mostly lung metastases. None were lost to follow-up. Four of them
were not evaluable for satellite tumorous cells because of inadequate and limited sampling of tumor periphery. Only 19 (52.8%, 19 out of 36) had available MRI-2 for late evaluation.
U
At baseline and early evaluation, diffuse IGP and stable or increased oedema were predictors
N
of DFS (HR=3.72, p=0.009 and HR=2.80, p=0.040, respectively).
A
At late evaluation, stable or increased oedema was significantly associated with a lower OS and DFS (HR=5.69, p=0.019 and HR=4.98, p=0.006, respectively), as well as change in
M
aponeurotic enhancement (HR=5.77, p=0.012 and HR=6.06, p=0.003, respectively). Interestingly, presence of satellite tumorous cells on surgical specimen was predictive of
TE D
lower OS and DFS at univariate analysis (HR=10.47 (1.35-81.5), p=0.005 and HR=4.64 (1.30-16.5), p=0.010, respectively – Fig. 2).
EP
3.5. Association between histological infiltrative pattern and histological response Good histological response and absence of satellite tumorous cells were not systematically associated as 4 good responders demonstrated satellite tumorous cells on surgical specimen.
CC
Three out of these 4 patients showed local (2 out of 4) and/or distant relapse in lung (2 out of 4) during follow-up and two of them died of disease 17 and 44.5 months after surgery.
A
Figure 3 illustrates one of these 4 cases. Figure 4 illustrate an opposite case of a living good responder without satellite tumorous cells with complete regression of peripheral anomalies on MRI. We also identified 4 patients with good responses and no satellite tumorous cells who demonstrated stable or slight increase of surrounding oedema and contrast-enhanced oedema. Radio-pathological correlations demonstrated a halo of inflammatory cells and activated myofibroblasts around a fibrotic capsule (Fig. 5).
4. DISCUSSION Surrounding tissues of STS demonstrate various aspects at baseline and during treatments but their relevance to predict histological response and prognosis is poorly understood. In addition, a better characterization of viable satellite tumorous cells beyond apparent tumorous borders is needed to delineate accurately surgical margins and radiotherapy fields. Overall, our results indicate that diffuse IGP on baseline MRI, poorly defined margins at early and late evaluation, as well as stable or increased contrast-enhanced oedema could help
IP T
to predict satellite tumorous cells on surgical specimen. Furthermore, changes in
surrounding oedema and in enhancement of this oedema may help to anticipate histological
SC R
response. Finally, diffuse IGP on baseline MRI and early change in surrounding oedema may predict of DFS.
Our results confirm those observed by Hanna et al. [25], that oedema would be associated with responses to NAC. Our results at baseline are also in accordance with studies that
U
showed associations between MRI IGP and histological IGP, as well as higher risk of poorer
N
DFS with diffuse IGP on MRI [7, 9]. Our results deepen these findings by investigating the
A
predictive value of early and late changes in surrounding tissues during NAC. In addition, we observed that therapy-related changes may dissociate with a good response within
M
tumour bulk while viable tumorous cells still persist in tissue periphery. Most patients of our cohort with this pattern of response displayed distant and local recurrences (3 of 4). We
TE D
hypothesize that tumour periphery represent in such cases the active invasion front of tumours, providing important insights regarding recurrence risk. This finding needs to be assessed on a larger cohort.
EP
If association between histological IGP and higher local recurrence rates is already well established [7, 31], few studies have investigated links between histological IGP, imaging of STS periphery and distant relapses [7], and none in the setting of NAC. The coexistence of
CC
Oedema, contrast-enhanced oedema, diffuse IGP on MRI may reflect a particular propensity of the tumour to metastasize.
A
The ability to anticipate poor responses during NAC course may help to adjust treatment. Likewise, MRI findings suspecting the presence of satellite tumorous cells in tumour periphery at last evaluation could help surgeons and radiotherapists to adjust their procedure (larger surgical margins, larger radiation fields, systemic adjuvant treatments) or advocate for closer monitoring at the end of treatments. Inter-observer agreements were moderate for most qualitative MRI features although consensus was easily reached between the two experts radiologists. This indicates that visual
11 assessment of surrounding tissue is complex and partly subjective. The retrospective nature of the study has probably biased the reproducibility because oedema was assessed on different MRI sequences, proton density -WI and T2-WI, with different fat suppression techniques which may biased the sensitivity to detect oedema [32, 33]. Similarly, contrast uptake of the surrounding oedema was estimated either on 3D gradient recalled echo or on 2D turbo spin echo fat-sat T1-WI, which may have different contrast-to-noise and signal-tonoise ratio, especially at 1.5 Tesla [34, 35]. It should be noted that focal or diffuse IGP on
IP T
MRI, contrast-enhanced oedema and tail sign were frequently found concomitantly in
patients. Even if we suggested strict definitions of each MRI feature, the distinction was
SC R
sometimes difficult to assess. This may also contribute to moderate reproducibility.
Consequently, further studies should investigate simplification of abnormal peripheryrelated MRI features.
MRI features in this study were all qualitative. Further studies should attempt to introduce
U
quantitative assessment of the surrounding oedema. In our experiment, basic measurements,
N
such as longest diameter of the oedema, were difficult to determine and poorly reproducible
A
because of the complex 3D nature of oedema. Next studies should propose homogeneous and standardized MRI protocols with large Field-of-view T2-WI and T1-WI after
M
Gadolinium-chelates injection with the same fat suppression method, same MRI contrast agent and same acquisition plan throughout the treatment.
TE D
Moreover, other imaging modalities may bring information regarding peritumoral tissue, such as DCE-MRI, DWI and 18FDG/PET-CT and should be added in further studies. Indeed, peritumoral tissues may harbour abnormal growth factors in addition to satellite tumorous
EP
cells resulting in slight quantitative differences in markers derived from their post-treatment. Interestingly, White et al. have demonstrated that satellite single or clusters of tumorous cells could be found beyond the tumour margins without correlation with the size of the
CC
tumour, oedema extension or contrast-enhanced oedema, in population of patients who were only treated by surgery. They found rare cases of satellite tumorous cells without associated
A
oedema [3]. One might hypothesize that more advanced imaging techniques could also identify these satellite tumorous cells without anomalies on conventional MRI, as well as being able to distinguish peritumoral enhancement due to inflammatory infiltrate from satellite tumorous cells. Our study has other limits. Only a limited numbers of patients were included and even a smaller number was used in the survival analysis. We decided to combine metastatic and distant relapses for the disease free survival because patient number and events would have
12 been too small otherwise for statistical analyses. We also simplified the histotypes for adjustment in the multivariate analyses into 2 categories: undifferentiated sarcomas vs. others. Undifferentiated sarcomas, which included undifferentiated pleomorphic sarcomas and myxofibrosarcomas, represent a category of tumours with complex genomic, poor prognosis and frequent abnormal MRI findings in their periphery [10-12]. Dichotomizing histotypes enabled to divide the population in 2 equivalent groups (n=27 vs. n=30) with a pathophysiological meaning. However, larger studies should follow the WHO classification
IP T
in order to better estimate the odds ratio (and hazard ratio) of the MRI features associated
with histological response and satellite tumors cells. To note, other studies that focused on
SC R
imaging biomarkers for STS did not exceed this population size, likely due to the rarity of these tumours [17, 21-24, 36].
Retrospective aspects of the study have also implications on the pathological analysis of the surgical specimen since STS periphery may have been under-sampled. Presence of satellite
U
tumorous cells may have been underestimated. However, we believe that variability in
N
treatments, surgical technics, pathological analyses were limited since patients were
A
managed by experts from a single sarcoma reference centre. Further studies should propose a homogeneous protocol for pathological management of surgical specimen. In that sense,
M
EORTC proposed guidelines for sarcoma pathologists [37] and similar initiatives should focus on tumour periphery. Other known or suspected predictors for OS and DFS were not
TE D
mentioned in the present study, such as C-reactive protein, vascular invasion, 18FDG/PETCT response or modified Choi criteria, and further studies should investigate multivariate models taking into account these co-variables.
EP
To conclude, this study emphasizes the importance of MRI analysis of STS tissue periphery as it may inform about histological response and remaining satellite tumorous cells distant from tumour borders. Further studies are needed to confirm these results and investigate
CC
prospectively STS periphery with standardized multimodal imaging and pathological
A
protocols.
Acknowledgements: The authors would like to thank Mrs Camille Martinerie for medical writing service.
Funding information: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
13
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Conflicts of interest: The authors declare no conflict of interest
14 REFERENCES [1] Honoré C, Méeus P, Stoeckle E, Bonvalot S, Soft tissue sarcoma in France in 2015: Epidemiology, classification and organization of clinical care, J Visc Surg. 152 (2015) 223– 230. [2] Zhao F, Ahlawat S, Farahani SJ, et al., Can MR imaging be used to predict tumor grade in soft-tissue sarcoma? Radiology. 272 (2014) 192–201.
IP T
[3] White LM, Wunder JS, Bell RS, et al., Histologic assessment of peritumoral edema in soft tissue sarcoma, Int J Radiat Oncol Biol Phys. 61 (2005) 1439–1445. [4] Panicek DM, Schwartz LH, Soft Tissue Edema Around Musculoskeletal Sarcomas at Magnetic Resonance Imaging, Sarcoma. 1 (1997) 189–191.
SC R
[5] Carneiro A, Bendahl P-O, Engellau J, et al., A prognostic model for soft tissue sarcoma of the extremities and trunk wall based on size, vascular invasion, necrosis, and growth pattern, Cancer. 117 (2011) 1279–1287.
U
[6] Sampo M, Tarkkanen M, Tukiainen E, et al., A web-based prognostic tool for extremity and trunk wall soft tissue sarcomas and its external validation, Br J Cancer. 106 (2012) 1076–1082.
A
N
[7] Fernebro J, Wiklund M, Jonsson K, et al., Focus on the Tumour Periphery in MRI Evaluation of Soft Tissue Sarcoma: Infiltrative Growth Signifies Poor Prognosis, Sarcoma. 2006 (2006) 21251.
TE D
M
[8] Park JW, Kim H-S, Lee C, Yoo HJ, Yun JY, Han I, Preoperative Factors Associated with Infiltrative Histologic Growth Patterns in Extremity Soft Tissue Sarcoma, Sarcoma. 2017 (2017) 5419394. [9] Nakamura T, Matsumine A, Matsubara T, et al., Infiltrative tumor growth patterns on magnetic resonance imaging associated with systemic inflammation and oncological outcome in patients with high-grade soft-tissue sarcoma, PLoS ONE. 12 (2017).
EP
[10] Kikuta K, Kubota D, Yoshida A, et al., An analysis of factors related to the tail-like pattern of myxofibrosarcoma seen on MRI, Skeletal Radiol. 44 (2015) 55–62.
CC
[11] Lefkowitz RA, Landa J, Hwang S, et al., Myxofibrosarcoma: prevalence and diagnostic value of the “tail sign” on magnetic resonance imaging, Skeletal Radiol. 42 (2013) 809–818.
A
[12] Yoo HJ, Hong SH, Kang Y, et al., MR imaging of myxofibrosarcoma and undifferentiated sarcoma with emphasis on tail sign; diagnostic and prognostic value, Eur Radiol. 24 (2014) 1749–1757. [13] Issels RD, Lindner LH, Verweij J, et al., Neo-adjuvant chemotherapy alone or with regional hyperthermia for localised high-risk soft-tissue sarcoma: a randomised phase 3 multicentre study, Lancet Oncol. 11 (2010) 561–570. [14] Saponara M, Stacchiotti S, Casali PG, Gronchi A, (Neo)adjuvant treatment in localised soft tissue sarcoma: The unsolved affair, Eur J Cancer. 70 (2017) 1–11.
15 [15] Pasquali S, Gronchi A, Neoadjuvant chemotherapy in soft tissue sarcomas: latest evidence and clinical implications, Ther Adv Med Oncol. 9 (2017) 415–429. [16] Gronchi A, Ferrari S, Quagliuolo V, et al., Histotype-tailored neoadjuvant chemotherapy versus standard chemotherapy in patients with high-risk soft-tissue sarcomas (ISG-STS 1001): an international, open-label, randomised, controlled, phase 3, multicentre trial, Lancet Oncol. 18 (2017) 812–822.
IP T
[17] Stacchiotti S, Collini P, Messina A, et al.. High-grade soft-tissue sarcomas: tumor response assessment--pilot study to assess the correlation between radiologic and pathologic response by using RECIST and Choi criteria, Radiology. 251 (2009) 447–456.
SC R
[18] Stacchiotti S, Verderio P, Messina A, et al., Tumor response assessment by modified Choi criteria in localized high-risk soft tissue sarcoma treated with chemotherapy, Cancer. 118 (2012) 5857–5866.
U
[19] Benz MR, Czernin J, Allen-Auerbach MS, et al., FDG-PET/CT imaging predicts histopathologic treatment responses after the initial cycle of neoadjuvant chemotherapy in high-grade soft-tissue sarcomas, Clin Cancer Res Off J Am Assoc Cancer Res. 15 (2009) 2856–2863.
A
N
[20] Benz MR, Allen-Auerbach MS, Eilber FC, et al., Combined assessment of metabolic and volumetric changes for assessment of tumor response in patients with soft-tissue sarcomas, J Nucl Med Off Publ Soc Nucl Med. 49 (2008) 1579–1584.
M
[21] Huang W, Beckett BR, Tudorica A, et al., Evaluation of Soft Tissue Sarcoma Response to Preoperative Chemoradiotherapy Using Dynamic Contrast-Enhanced Magnetic Resonance Imaging, Tomogr J Imaging Res. 2 (2016) 308–316.
TE D
[22] Xia W, Yan Z, Gao X, Volume fractions of DCE-MRI parameter as early predictor of histologic response in soft tissue sarcoma: A feasibility study, Eur J Radiol. 95 (2017) 228– 235.
EP
[23] Soldatos T, Ahlawat S, Montgomery E, Chalian M, Jacobs MA, Fayad LM, Multiparametric assessment of treatment response in high-grade soft-tissue sarcomas with anatomic and functional MR imaging sequences, Radiology. 278 (2015) 831–840.
CC
[24] Meyer JM, Perlewitz KS, Hayden JB, et al., Phase I trial of preoperative chemoradiation plus sorafenib for high-risk extremity soft tissue sarcomas with dynamic contrast-enhanced MRI correlates, Clin Cancer Res Off J Am Assoc Cancer Res. 19 (2013) 6902–6911.
A
[25] Hanna SL, Fletcher BD, Parham DM, Bugg MF, Muscle edema in musculoskeletal tumors: MR imaging characteristics and clinical significance, J Magn Reson Imaging JMRI. 1 (1991) 441–449. [26] Bahig H, Roberge D, Bosch W, et al., Agreement among RTOG sarcoma radiation oncologists in contouring suspicious peritumoral edema for preoperative radiation therapy of soft tissue sarcoma of the extremity, Int J Radiat Oncol Biol Phys. 86 (2013) 298–303.
16 [27] Trojani M, Contesso G, Coindre JM, et al., Soft-tissue sarcomas of adults; study of pathological prognostic variables and definition of a histopathological grading system, Int J Cancer. 33 (1984) 37–42. [28] Eisenhauer EA, Therasse P, Bogaerts J, et al., New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1), Eur J Cancer. 45 (2009) 228–247.
IP T
[29] Cousin S, Crombé A, Italiano A, et al., Clinical, radiological and genetic features associated with the histopathologic response to noadjuvant chemotherapy (NAC) and outcomes in locally advanced soft tissue sarcoma (STS) patients, Journal of Clinical Oncology. 35:15_suppl (2017) 11014-11014. [30] Landis JR, Koch GG, The measurement of observer agreement for categorical data. Biometrics. 33 (1977) 159–174.
SC R
[31] Manoso MW, Pratt J, Healey JH, Boland PJ, Athanasian EA, Infiltrative MRI pattern and incomplete initial surgery compromise local control of myxofibrosarcoma, Clin Orthop. 450 (2006) 89–94.
N
U
[32] Delfaut EM, Beltran J, Johnson G, Rousseau J, Marchandise X, Cotten A, Fat suppression in MR imaging: techniques and pitfalls, Radiogr Rev Publ Radiol Soc N Am Inc. 19 (1999) 373–382.
A
[33] Bley TA, Wieben O, François CJ, Brittain JH, Reeder SB, Fat and water magnetic resonance imaging, J Magn Reson Imaging JMRI. 31 (2010) 4–18.
M
[34] Hargreaves BA, Rapid gradient-echo imaging, J Magn Reson Imaging JMRI. 36 (2012) 1300–1313.
TE D
[35] Zur Y, Wood ML, Neuringer LJ, Spoiling of transverse magnetization in steady-state sequences, Magn Reson Med. 21 (1991) 251–263.
EP
[36] Favinger JL, Hippe DS, Davidson DJ, et al., Soft Tissue Sarcoma Response to Two Cycles of Neoadjuvant Chemotherapy: A Multireader Analysis of MRI Findings and Agreement with RECIST Criteria and Change in SUVmax, Acad Radiol. 25 (2018) 470– 475.
A
CC
[37] Wardelmann E, Haas RL, Bovée JVMG, et al., Evaluation of response after neoadjuvant treatment in soft tissue sarcomas; the European Organization for Research and Treatment of Cancer–Soft Tissue and Bone Sarcoma Group (EORTC–STBSG) recommendations for pathological examination and reporting, Eur J Cancer. 53 (2016) 84– 95.
17 LEGENDS
SC R
IP T
Figure 1. MRI features depicting surrounding tissue of soft-tissue sarcoma. (A) ‘Pushing-type’ growth pattern (: entirely well-defined tumour) in a grade III undifferentiated round cells sarcoma of the thigh. (B) ‘Focal-type’ infiltrative growth pattern (: irregular borders and infiltration of surrounding tissue <25% of tumour circumference, white arrow) in a grade III pleomorphic liposarcoma of the thigh. (C) ‘Diffuse-type’ infiltrative growth pattern (: irregular borders and infiltration of surrounding tissue ≥25% of tumour circumference) in a grade III undifferentiated epitheloid sarcoma of the arm. (D) Aponeurotic enhancement of ≥2mm thickness (‘tail sign’, white arrows) in a grade III myxofibrosarcoma of the arm. (E) Surrounding oedema (white arrow) without contrast enhancement in a grade III undifferentiated pleomorphic sarcoma of the psoas muscle. (F) Contrast enhancement of a surrounding oedema (white arrow), without mass effect in a grade III dedifferentiated liposarcomas of the thigh. T2: T2 weighted imaging, FS T1+: Fatsuppressed T1 weighted imaging after Gadolinium-chelates injection, FS T2: T2 weighted imaging with fat suppression.
A
N
U
Figure 2. Kaplan-Meier curves of MRI and pathological features to predict diseasefree survival. Significant separation was determined by log rank test, *: p<0.05, **: p<0.005
A
CC
EP
TE D
M
Figure 3. Radio-pathological correlation of a good histological responder with persistent satellite tumorous cells after neoadjuvant chemotherapy. Patient presented with a deeply-located grade III leiomyosarcoma of the left quadriceps femori. (A) MRI at baseline and (B) early evaluation: during neoadjuvant chemotherapy, tumor demonstrated strong decrease of T2 signal intensity compatible with a fibrotic response (white arrow), together with foci of necrosis. However, surrounding oedema rather tended to increase (white arrow-head) and still displayed marked enhancement after Gadolinium-chelates injection (lined white arrow-head). (C) Pathological analysis of the surgical specimen confirmed a good response with 5% stainable cells, 85% fibrosis (Δ), 10% necrosis and demonstrated spans of fibrosis (°) between muscular bundles (§) but magnification (black frame) showed single and clusters of satellite tumorous cells beyond the tumour borders (black arrow). Follow-up demonstrated lung metastatic relapse 42 months after surgery and patient died of disease 4 months later. FS PD: Fat-Sat proton density weighted imaging, T2: T2 weighted imaging, FS T1+: Fat-Sat T1 weighted-imaging after Gadolinium-chelates injection. Figure 4. Example of a good responder without satellite tumour cells on surgical specimen: classical evolution of the surrounding tissue during neoadjuvant chemotherapy. Patient presented with a deep and superficial grade III undifferentiated pleomorphic sarcoma of the thigh. (A) Baseline MRI showed a diffuse infiltrative growth pattern with poorly defined margin on >25% tumour circumference, associated with surrounding oedema (white arrow) that enhanced after Gadolinium-chelates injection, and
18 with thick aponeurotic enhancement spreading along the superficial fascia (white arrow head). After 2 cycles of chemotherapy (B), these peripheral anomalies decreased in an unequivocal manner and >50% of tumour volume demonstrated fibro-necrotic changes (white asterix). (C) Pathological analysis confirmed a good response. Nine years later, patient is alive in remission. T2: T2-weighted imaging, FS T1+: Fat-Sat T1-weighted imaging after Gadolinium-chelates injection.
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Figure 5. Example of a good responder without satellite tumorous cells on surgical specimen: atypical evolution of the surrounding tissue during neoadjuvant chemotherapy. Patient presented with a superficial and aponeurotic grade III myxofibrosarcoma of the upper limb. (A) Tumour initially showed a focal infiltrative growth pattern. At early evaluation (B) and late evaluation (C), oedema tended to increase (white arrow), with additional enhancement in the subcutaneous fat and increased aponeurotic enhancement (white arrowhead). However, tumour demonstrated extensive fibro-necrotic changes suggestive of a good response with occurrence of fibrotic capsule (dashed white arrow). These findings were still observed at late evaluation. (D) Pathological analysis confirmed a good histologic response and found a thick fibrotic capsule (°) with two fronts of inflammatory cells (*) made of lymphocytes, histiocytes, inside and outside, as well as activated fibroblasts on magnification (black frame). No satellite tumorous cell was found and we concluded that the increased surrounding anomalies were due to an inflammatory reaction.
EP
CC
A TE D
IP T
SC R
U
N
A
M
19
EP
CC
A TE D
IP T
SC R
U
N
A
M
20
EP
CC
A TE D
IP T
SC R
U
N
A
M
21
EP
CC
A TE D
IP T
SC R
U
N
A
M
22
23 Table 1. Population study.
21 (36.8) 36 (63.2)
Age (in years)
58 (31-77)
Histotypes Undifferentiated pleomorphic sarcoma Myxofibrosarcoma Rhabdomyosarcoma Leiomyosarcoma M/RC-LPS Dedifferentiated liposarcomas Pleomorphic liposarcoma Synovial sarcoma MPNST
22 (38.6) 5 (8.8) 8 (14) 5 (8.8) 6 (10.5) 1 (1.8) 3 (5.3) 6 (10.5) 1 (1.8)
M
A
2 (3.5) 41 (71.9) 14 (24.6)
TE D
Situation Shoulder Girdle Upper Limb Trunk Wall Pelvic Girdle Lower Limb
N
Depth Superficial Deep Deep + Superficial
LD at baseline (in mm)
EP
U
Gender Female Male
IP T
No of patients
SC R
Characteristics
Time last cycle to Surgery (in days)
6 (10.5) 6 (10.5) 8 (14) 2 (3.5) 35 (61.4) 102 (39-260) 31 (17-64)
A
CC
NOTE - LD : longest diameter. M/RC-LPS : Myxoid / round cells liposarcoma, MPNST : malignant peripheral nerve sheath tumor. Results are given as numbers of patients with percentage in parentheses, except for age, longest diameter at baseline and time from last cycle to surgery, which are given as median and range in parentheses.
24 Table 2. Inter-observer agreement of the MRI features and their changes from baseline (MRI-0) to early (MRI-1) and pre-surgery evaluation (MRI-2). Table 2. Inter-observer agreement of the MRI features and their changes from baseline MRI features
κ = 0.597, (0.217-0.877)
p=0.000
Contrast-enhanced oedema
κ = 0.624, (0.391-0.857)
p=0.000
Aponeurotic enhancement
κ = 0.602, (0.404-0.800)
p=0.000
Bone, vessel or nerve invasion
κ = 0.855, (0.720-0.990)
p=0.000
wκ = 0.427, (0.239-0.615)
p=0.000
Change in margin definition
κ = 0.335, (0.119-0.551)
p=0.003
Change in oedema
κ = 0.527, (0.305-0.749)
p=0.000
Change in contrast-enhanced oedema
κ = 0.553, (0.349-0.757)
p=0.000
Change in Aponeurotic enhancement
κ = 0.304, (0.049-0.559)
p=0.021
Change in margin definition
κ = 0.243, (0.027-0.459)
p=0.049
Change in oedema
κ = 0.634, (0.372-0.897)
Change in aponeurotic enhancement
A
Change in contrast-enhanced oedema
p=0.000
κ = 0.299, (0.001-0.597)
p=0.042
κ = 0.211, (0-0.556)
p=0.183
N
MRI-2 (N=36)
SC R
Growth pattern MRI-1 (N=57)
IP T
Oedema
U
MRI-0 (N=57)
Inter-Observer Agreement (CI95%, p-value)
M
(MRI-0) to early (MRI-1) and pre-surgery evaluation (MRI-2).
A
CC
EP
TE D
NOTE – Abbreviations : CI95% : 95% confidence interval, κ : Cohen’s kappa ; LD : longest diameter, wκ : weighted kappa.
25
Table 3. Associations between pathological findings on surgical specimen and MRI features at baseline (MRI-0), early (MRI-1) and late evaluations (MRI-2). Histological Response Good
Poor
OR (CI95%)
Satellite tumor cells p-value
No
Yes
OR (CI95%)
p-value
MRI-0 Oedema
No
1
5
_
Yes
14
37
0.53 (0.06-4.93)
No
4
12
_
3
3
_
25
22
0.88 (0.16-4.82)
10
5
_
18
20
2.22 (0.64-7.74)
15
10
11
15
16
18
12
7
8
1
_
10
7
9.60 (1.10-83.40) 1
10
17
3.82 (1.122-11.98) 2
0
0
_
1
2
_
24
21
_
1.000
Aponeurotic enhancement
1.000 Yes
11
30
0.91 (0.24-3.42)
No
7
19
_
0.237
1.000
MRI growth pattern
7
22
1.16 (0.34-3.90)
No
10
25
_
Yes
5
17
1.36 (0.39-4.69)
Pushing
1
8
_
0.761
1
Focal infiltration
4
14
0.30 (0.03-2.66)
Diffuse
10
20
0.46 (0.13-1.56) 2
CR
0
0
_
PR
1
1
_
Oedema 3
Contrast enhanced oedema Aponeurotic enhancement 3
PD
1
4
Better delimited
4
8
Stable or worst
11
Decrease or absent
9
Increase or stable
6
Decrease or absent
11
Contrast-enhanced oedema Aponeurotic enhancement
11
_
30
4.09 (1.18-14.18)
15
_ 4.95 (1.34-18.29)
Decrease or absent
11
23
_
_
0.030 *
0.017 *
3
2
_
11
1
_
17
24
15.53 (1.83-131.9)
13
6
15
18
2.60 (0.79-8.51)
18
6
_
10
19
5.70 (1.72-18.92)
20
12
_
19
3.03 (0.74-12.45)
7
13
3.09 (0.97-9.92)
CR
0
0
_
0
0
_
3
3
_
3
2
_
0.153
0.326
SD
6
22
_
14
12
_
PD
0
3
_
0
2
_
Better delimited
5
7
10
2
_
7
14
10.00 (1.71-56.63)
21
3.75 (0.78-17.99)
Decrease or absent
9
13
_
Increase or stable
0
15
0 (0-0.47)
Decrease or absent
8
15
_
0.005 **
13
6
_
4
10
5.42 (1.20-24.52)
13
7
_
4
9
4.18 (0.94-18.16)
15
9
_
2
7
5.833 (0.988-34.436)
0.112 Increase or stable
1
13
6.93 (0.76-63.05)
Decrease or absent
7
20
_
2
8
1.40 (0.24-8.24)
0.010 *
0.037 *
0.080
1.000 Increase or stable
0.005 **
0.087
3
4
0.003 **
0.152
Increase or stable
Stable or worst
0.022 *
0.754
0.205
0.116
0.390
0.52 (0.16-1.64)
0.713
1.55 (0.39-6.14)
27
EP CC
A
Oedema
_
4
PR
Margin definition
_
0.267
2.05 (0.670-6.24)
0.710
_
Increase or stable
MRI-0 to MRI-2 RECIST 1.1
34
N
37
A
13
M
Margin definition
SD
TE D
RECIST 1.1
0.370
U
MRI-0 to MRI-1
_
SC R
Bone, vessel or nerve invasion
Yes
IP T
Contrast-enhanced oedema
1.000
0.057
NOTE – Abbreviations: CI95%: 95% confidence interval, CR : complete response, OR: odds ratio, PD : progressive disease, PR : partial response, SD : stable disease. MR features with 3 values were dummying as follows : 1 : Pushing growth pattern versus presence of infiltration, 2 : pushing or focal infiltration versus infiltrating. 3 : At early evaluation, one patient was removed for the assessment of
26 change in contrast-enhanced oedema and one another patient for aponeurotic enhancement as STIR T2-WI and post-Gd FS T1-WI did not capture the whole abnormal findings, thus the statistical analysis was made on 56 patients instead of 57. * : p<0.05, **, p<0.005.
Table 4. Multivariate analysis of the MRI features associated with pathological findings Histological Response OR (CI95%)
Satellite tumor cells
p-value
OR (CI95%)
p-value
MRI-0 Pushing or Focal
_
Diffuse
_
better delimited
_
_
_
3.52 (1.06-11.67)
MRI-0 to MRI-1
Oedema
Stable or worst
_
Decrease or absent
_
Increase or stable Contrast-enhanced oedema
_ _
_
0.011 *
6.87 (1.56-30.33)
Decrease or absent
_
Increase or stable
8.06 (1.74-37.32)
19.06 (1.97-185.6)
SC R
Margin definition
0.039 *
IP T
MRI growth pattern
_
0.011 *
_
_
0.008 *
5.59 (1.54-20.36)
0.009 *
A
CC
EP
TE D
M
A
N
U
NOTE – Abbreviations: CI95%: 95% confidence interval, OR: odds ratio. Results were adjusted for histotype (myxofibrosarcoma or undifferentiated sarcoma versus others), age (< or ≥ median age = 58 years old) and size at baseline (< or ≥ median size = 102 mm). *: p<0.05.
27 Table 5: Univariate survival analysis of the MRI features evaluated at baseline (MRI-0), at early evaluation (MRI-0 to MRI-1) and late evaluation (MRI-0 to MRI-2) Overall Survival Survival (CI95%)
HR (CI95%)
No Yes No Yes No Yes No Yes Pushing or focal Diffuse
94.6 (51.1-138.2) 70.3 (53-87.6) 51 (33.8-28.3) 82.1 (61-103.2) 66.9 (36.1-97.8) 48.7 (34-63.5) 77 (50.7-103.2) 47.4 (34.8-60) 94.9 (69.3-120.6) 45.8 (32.6-59)
_ 1.51 (0.19-12.13) _ 0.89 (0.25-3.25) _ 1.22 (0.41-3.66) _ 1.16 (0.38-3.54) _ 2.18 (0.67-7.11)
Objective response No response Better delimited Stable or worst Decrease or absent Increase or stable Decrease or absent Increase or stable Decrease or absent Increase or stable
34.5 (19.6-49.5) 76.9 (56-97.8) _£ 67.1 (44.7-89.5) 79.7 (53-106.4) 66.1 (41.4-90.9) 82.4 (57.3-107.6) 63.8 (38.4-89.3) 83.7 (53.7-114) 42.3 (24.6-60)
_ 0.73 (0.09-5.72) _ 28.99 (0.07-12437) _ 2.88 (0.79-10.51) _ 3.46 (0.94-12.68) _ 2.92 (0.94-9.01)
Objective response No response Better delimited Stable or worst Decrease or absent Increase or stable Decrease or absent Increase or stable Decrease or absent
34.5 (19.6-49.5) 39.5 (29.8-49.3) 86.6 (46.1-127.1) 38.9 (29.1-48.8) 93.3 (72.3-81) 25.5 (15-35.9) 85.7 (61.8-109.5) 30.9 (17-44.8) 84.5 (62.9-106)
_ 0.99 (0.12-8.09) _ 2.09 (0.26-17.15) _ 5.69 (1.13-28.77) _ 3.45 (0.81-14.73) _
Increase or stable
19.5 (11.4-127.6)
5.77 (1.24-26.91)
Disease-free Survival p-value
Survival (CI95%)
HR (CI95%)
84.3 (26.5-142) 48.1 (30.5-65.8) 32.1 (16.3-47.9) 57.8 (35.7-97.9) 56.3 (28.2-84.3) 37.6 (14-41.3) 68.9 (36.7-90.9) 30.5 (16.6-44.3) 83.2 (53.6-113) 23.3 (12.2-33.8)
_ 2.02 (0.27-15.24) _ 0.96 (0.32-2.89) _ 1.55 (0.62-3.89) _ 1.47 (0.59-3.63) _ 3.72 (1.31-10.54)
p-value
Margin definition Oedema Contrast-enhanced oedema Aponeurotic enhancement
MRI-0 to MRI-2 RECIST1.1 Margin definition Oedema Contrast-enhanced oedema Aponeurotic enhancement
0.787 0.184
SC R
MRI-0 to MRI-1 RECIST1.1
0.721
0.763
0.080 0.095
U
Growth pattern
TE D
bone, vessel or nerve invasion
0.861
0.052
N
Aponeurotic enhancement
0.697
0.052
A
Contrast-enhanced oedema
M
Oedema
IP T
MRI-0
0.993 0.481 0.019 * 0.077 0.012 *
19.3 (0-43.6) 57.9 (37.2-78.6) _£ 43.7 (24.2-63.2) 40.4 (43.8-97) 35.8 (13.6-58) 60.4 (36.1-84.8) 47.7 (22-73.4) 65.9 (93.7-92.1) 26.9 (9.2-44.6)
_ 0.45 (0.10-1.99) _ 30.74 (0.25-3788) _ 2.80 (1.01-7.84) _ 1.85 (0.72-4.73) _ 2.37 (0.95-5.89)
19.3 (0-43.6) 28.6 (18-39.3) 85 (41.8-128) 24.7 (14.8-34.6) 67.1 (40.2-94) 14.2 (5-23.5) 59.2 (31.8-86.6) 1.7 (5.7-29.3) 57.6 (33.5-81.7)
_ 0.52 (0.11-2.43) _ 5.85 (0.69-43.42) _ 4.98 (1.42-17.40) _ 2.93 (0.94-9.09) _
9.2 (1.8-16.6)
6.06 (1.57-23.44)
A
CC
EP
NOTE – Abbreviations: CI95%: 95% confidence intervals, HR: hazard ratio, LD: longest diameter. Mean survivals are given in months. HR and survivals are given with CI95%. *: p<0.05, **: p<0.005, £: not calculable as only 4 patients displayed this pattern and none of them died or had an event related to disease.
0.486 0.938 0.340 0.404 0.009 *
0.279 0.08 0.040 * 0.191 0.056
0.398 0.073 0.006 * 0.054 0.003 **