- Email: [email protected]
deep-seated ‘lipomas’: incidence, predictors and clinical significance
Pathology (April 2016) 48(3), pp. 203–209
A N ATO M I C A L PAT H O L O G Y
MDM2/CDK4 gene amplification in large/deep-seated ‘lipomas’: incidence, predictors and clinical significance DANIEL D. WONG1,2, IRENE C. LOW1, JOANNE PEVERALL3, PETER D. ROBBINS1, DOMINIC V. SPAGNOLO1, ROBERT NAIRN4, RICHARD L. CAREY-SMITH5 AND DAVID WOOD5 1
Department of Anatomical Pathology, PathWest Laboratory Medicine, QEII Medical Centre, Nedlands, 2School of Pathology and Laboratory Medicine, University of Western Australia, Crawley, 3Department of Diagnostic Genomics, PathWest Laboratory Medicine, QEII Medical Centre, Nedlands, 4Department of Radiology, Sir Charles Gairdner Hospital, Nedlands, and 5Perth Orthopaedic Institute, Nedlands, WA, Australia
Summary This study of 140 cases assessed the incidence of MDM2/ CDK4 gene amplification in lipomatous neoplasms with histological features of a lipoma but which were of clinical concern due to large size (50 mm) and/or deep-seated (subfascial) location. Univariate and multivariate statistical analyses were used to identify clinical, radiological and pathological predictors of gene amplification. Differences in local recurrence rates between amplified and nonamplified cases were assessed using survival analysis. The findings indicate that the incidence of MDM2/CDK4 amplification in this setting is low at 5% (95%CI 1.4–8.6%). Variables associated with amplification on univariate analysis were tumour site (thigh, p = 0.004), size (>100 mm, p = 0.033) and presence of equivocal atypia (p = 0.001). Independent predictors on multivariate analysis were size (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and presence of equivocal atypia (OR 12.5, 95%CI 1.9–80.3, p = 0.008). There was no significant difference in local recurrence rates between amplified and non-amplified cases (p = 0.461) based on a median follow-up time of 31 months. Assessment for MDM2/CDK4 amplification, therefore, should be considered in ‘lipomas’ which are >100 mm in size, show equivocal atypia and arise in the thigh. However, the clinical significance of gene amplification in this setting is unclear and requires confirmation in larger studies. Key words: MDM2; CDK4; amplification; lipoma; liposarcoma. Received 16 October, revised 26 November, accepted 6 December 2015 Available online 2 March 2016
may be useful to help aid in this distinction. Specifically, well differentiated liposarcomas are cytogenetically characterised by giant marker and ring chromosomes involving the chromosomal region 12q13-15.1,2 MDM2 (murine double minute 2) and CDK4 (cyclin dependent kinase 4) are two of several genes known to be amplified in this region and detection of these genetic changes has demonstrated high sensitivity and specificity for the histological diagnosis of well differentiated liposarcoma.3–12 What is less certain, however, is whether testing for MDM2/CDK4 gene amplification is indicated in bland lipomatous neoplasms with histological features of a lipoma but which are concerning due to large size and/or a deepseated (subfascial) location. Whilst it would seem that this issue has been previously addressed, most studies which have examined these alterations in lipomatous neoplasms have only included small numbers of lipomas in their validation series.3–12 Equally uncertain is the clinical significance of demonstrating MDM2/CDK4 amplification in this setting. Although it is commonly assumed, extrapolating from early studies, that these molecular changes warrant a diagnosis of liposarcoma irrespective of the microscopic findings,6,13 there are presently few published empirical data to support this assumption. The aim of this study, therefore, was to address some of the issues related to this subject: first, to determine the incidence of MDM2/CDK4 amplification in a series of apparent ‘lipomas’ of large size and/or deep-seated (subfascial) location; second, to identify clinical, radiological and pathological factors which might predict for these alterations and help avoid unnecessary costly and time consuming molecular tests; and finally, to investigate whether the finding of gene amplification in this setting is associated with a more aggressive clinical course, warranting classification separate from simple lipomas.
INTRODUCTION
MATERIALS AND METHODS
Lipomas are among the most common soft tissue neoplasms encountered by surgical pathologists. Separating lipomas from well differentiated liposarcomas is usually not difficult in routine practice based on their microscopic appearances. In more problematic cases, assessment of genetic alterations
Well differentiated lipomatous neoplasms submitted for assessment of MDM2/CDK4 amplification by fluorescence in situ hybridisation (FISH) between January 2009 and August 2015 were retrospectively identified from our laboratory database. Inclusion criteria for this study were: (1) largest macroscopic dimension 50 mm and/or deep-seated (subfascial) location; (2) absence of diagnostic atypical morphological features which would lead to a
Print ISSN 0031-3025/Online ISSN 1465-3931 Crown Copyright © 2016 Published by Elsevier B.V. on behalf of The Royal College of Pathologists of Australasia. All rights reserved. DOI: http://dx.doi.org/10.1016/j.pathol.2016.02.007
204
WONG et al.
histological diagnosis of well-differentiated liposarcoma (namely, overtly atypical stromal cells showing enlarged and hyperchromatic nuclei with or without atypical lipoblasts). Retroperitoneal and pelvic lipomatous neoplasms, which have a marked predisposition for local recurrence even in the absence of atypical morphological features,14,15 were excluded. Cases with a prior diagnosis of liposarcoma and presenting with a recurrent mass were also excluded. Most tumours (n = 134, 95.7%) were reported by subspecialty soft tissue pathologists in our department (DDW, ICL, PDR and DVS). Haematoxylin and eosin (H&E) slides from all remaining cases were reviewed to confirm that the histological features satisfied the inclusion criteria for the study. Clinical and radiological data were obtained from pathology requisition forms and radiology reports. Variables considered included patient age, gender, tumour location, tissue plane involved (superficial, subfascial or intramuscular), tumour size (largest dimension) on imaging studies, imaging modality used [ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI)] and the presence of atypical imaging features (presence of thick septa, abnormal signal patterns and lesional enhancement following administration of contrast). The imaging diagnosis favoured by the reporting radiologist was noted, retrieved from the conclusion of radiology reports and in many cases, confirmed at our soft tissue multidisciplinary team meetings in which an experienced musculoskeletal radiologist was present. Follow-up information including time to first local recurrence as confirmed on surveillance imaging studies was obtained by interrogating the databases of radiological centres known to service our soft tissue surgeons who were involved in the management of the majority of the cases included in this study (n = 106, 75.7%). Pathological data considered included tumour size (largest dimension) on macroscopic examination, number of blocks submitted, the presence of unusual histological features (fat necrosis, inflammation, equivocal atypia in stromal cells falling short of that allowing a definite morphological diagnosis of liposarcoma) and the final histological diagnosis (including classification as a lipoma variant). For specimens which were received in multiple pieces, the maximum dimension of the largest piece was accepted as the best approximation for the overall lesion size (compared to aggregate dimensions which, in our experience, are less reproducible). To assess the adequacy of sampling, a block:size ratio was determined by dividing the number of blocks submitted for microscopic examination by the largest dimension of the tumour size in cm. Interphase FISH analysis to assess for the presence of MDM2/CDK4 amplification was performed on representative formalin fixed, paraffin embedded sections using in-house DNA probes specific for the MDM2 gene at 12q15, CDK4 gene at 12q15 and chromosome 12 centromere (CEP12). Areas of interest were targeted by correlating with serial H&E stained sections. Two hundred successive nuclei were examined by two independent analysts (including JP) using the Applied Imaging CytoVision image capture system (Leica Biosystems, Germany). MDM2:CEP12 and CDK4:CEP12 signal ratios 2 in more than 15% of tumour cells were considered amplified. Signal ratios between 2 and 10 were classified as low level amplification and ratios >10 as high level amplification. Statistical comparisons of proportions and means were performed using Pearson’s chi-square test and Student’s independent samples t-test, respectively. Pearson’s correlation was used to assess the strength of the correlation between the largest dimension of lesions measured macroscopically compared to those measured on imaging studies. Multivariate logistic regression analysis was then performed to identify any independent predictors of MDM2/CDK4 amplification. The forward stepwise conditional method was used to introduce selected variables of clinical interest and those found to be significant on univariate analysis (including age, gender, tumour site, tissue plane involved, tumour size, presence of equivocal atypia and final radiological diagnosis) into the multivariate equation. Kaplan–Meier survival curves were used to analyse the time to first local recurrence according to MDM2/CDK4 amplification status and the difference between the curves was examined using the log-rank test. Statistical analyses were performed using SPSS for Windows Version 20.0 (IBM Corporation, USA) with the two-sided statistical significance level set at 5%. This study was approved by the Institutional Review Board (QA# 9748). Informed consent was waived as there was no direct patient involvement and
Pathology (2016), 48(3), April
the results of this retrospective study were not anticipated to alter clinical management.
RESULTS A total of 140 cases were identified from our database meeting the inclusion criteria for the study, including 83 (59.3%) males and 57 (40.7%) females. The mean and median ages were 56.5 years and 57 years, respectively (range 21–88 years). Specific sites of the tumours are detailed in Table 1. The lower limbs were involved in 50 (35.7%) cases, upper limbs in 47 (33.6%), trunk (chest wall, abdominal wall and back) in 41 (29.3%) and head/neck region in two (1.4%). Of those involving the extremities, 65 (50.4%) involved the left side and 64 (49.6%) involved the right. Up to 23 (16.4%) were situated within the subcutaneous fat, 54 (38.6%) intramuscular and the remaining 62 (44.3%) were deep/subfascial but extramuscular. Imaging data were available for 100 (71.4%) cases, of which 82 (82%) were assessed with MRI, 10 (10%) with CT and 8 (8%) with ultrasound. Of these, the preferred radiological diagnosis by the reporting radiologist was that of a simple lipoma in 78 (78%), whereas a diagnosis of a welldifferentiated liposarcoma was either favoured or could not be confidently excluded in the remaining 22 (22%), based purely on the imaging characteristics, irrespective of size and location. The frequency of atypical imaging features (the presence of thick septa, abnormal signal and abnormal enhancement following contrast administration) in the latter group is detailed in Table 2. Up to 124 (88.6%) specimens were received intact. In the remaining 16 (11.4%) which were received in multiple pieces, tumour size was approximated by taking the maximum dimension of the largest piece. Mean and median tumour size on macroscopic examination was 95 mm and 85 mm, respectively (range 20–275 mm). Tumour dimensions as measured on imaging studies were available in 96 (68.6%) cases with a mean and median size of 102 mm and 90 mm, respectively (range 18–390 mm). Correlation between the macroscopic and radiologically determined dimensions for this subset was strong [r = 0.74, 95% confidence interval (CI) 0.64–0.82, p < 0.001; Fig. 1]. The average number of paraffin blocks submitted per centimetre of the maximum tumour dimension was 1.2 ± 0.4 (range 0.4–3.0). Microscopic examination of all 140 cases revealed features consistent with a simple lipoma in 119 (85%). Specifically, the lesions comprised solid sheets and large lobules of mature adipose tissue often forming circumscribed but encapsulated masses. Occasional thin, intervening, paucicellular fibrous septa containing isolated bland spindled fibroblastic cells were seen in most tumours (Fig. 2). Adipocytes displayed no more than mild variability in size and shape without the presence of well-developed univacuolated and multivacuolated lipoblasts. Fat necrosis featured in five (4.2%) cases whilst a patchy lymphocytic infiltrate was noted in two (1.7%). Ten (7.1%) showed more frequent and thicker paucicellular bands of fibrosis and were classified as fibrolipomas. Spindle cell lipomas accounted for eight (5.7%) cases showing the well-recognised features of a lipomatous neoplasm with plump, uniform CD34+ spindle cells set in a variably fibrous and myxoid stroma, with coarse collagen fibrils and mast cells. Other variants included osteolipomas [lipoma with metaplastic ossification, n = 2 (1.4%)] and a
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
Table 1
Clinicopathological characteristics of all included cases
Variable
Statistic
Mean age, years (range) (n = 140) Sex, male:female (n = 140) Tumour site (n = 140) Upper arm Lower arm/hand Thigh Lower leg/foot Trunk Head/neck Tissue plane (n = 139) Subcutaneous Intramuscular Subfascial (extramuscular) Macroscopic tumour size (n = 139) <50 mm 50–99 mm 100–150 mm >150 mm Histological classification (n = 140) Simple lipoma Fibrolipoma Spindle cell lipoma Other subtypes
56.5 (21–88) 3:2 35 12 42 8 41 2
(25%) (8.5%) (30%) (5.8%) (29.3%) (1.4%)
23 (16.5%) 54 (38.8%) 62 (44.6%) 16 70 35 18
(11.5%) (50.4%) (25.2%) (12.9%)
119 10 8 3
(85%) (7.1%) (5.7%) (2.1%)
hibernoma [conventional type, n = 1 (0.7%)]. By definition, none of the cases showed atypical stromal cells with enlarged and hyperchromatic nuclei, diagnostic of well differentiated liposarcoma. Of those classified as a simple lipoma, additional features of note included the presence of equivocal atypia (isolated stromal cells showing mild nuclear enlargement without hyperchromasia, falling short of the diagnostic features to allow a histological diagnosis of welldifferentiated liposarcoma, Fig. 2) in 12 (10.1%). MDM2/CDK4 amplification was assessed by FISH in all 140 cases. The MDM2 gene was amplified in seven (5%, 95%CI 1.4–8.6%) cases, with a median MDM2/CEP12 ratio of 20 (range 2–60). Three showed low level and four showed high level amplification, respectively. CDK4 was amplified in four (2.9%, 95%CI 0.1–5.7%) cases, with a median CDK4/CEP12 ratio of 4.5 (range 2–30). Three showed low
Table 2
Summary of imaging findings
Variable Imaging modality (n = 100) MRI CT US Radiological tumour size (n = 97) <50 mm 50–99 mm 100–150 mm >150 mm Atypical imaging features (n = 100) Thick septa Lesional enhancement Abnormal (non-fat) signal Favoured diagnosis (n = 100) Simple lipoma Liposarcomaa
Statistic
82 (82%) 10 (10%) 8 (8%) 7 47 26 17
(5%) (33.6%) (18.6%) (12.1%)
14 (14%) 12 (12%) 5 (5%) 78 (78%) 22 (22%)
CT, computed tomography; MRI, magnetic resonance imaging; US, ultrasound. a
Liposarcoma favoured or could not be confidently excluded.
205
level and one showed high level amplification, respectively. MDM2 and CDK4 were co-amplified in four cases. The clinicopathological characteristics of amplified cases are summarised in Table 3. Univariate statistical analysis was performed to investigate the association of various clinical, radiological and pathological factors with MDM2/CDK4 amplification. The results of this analysis are detailed in Table 4. In summary, factors significantly associated with gene amplification were tumour site (p = 0.004), tumour size (both macroscopic and on imaging studies; p = 0.033 and p = 0.013, respectively) and histological features (presence of equivocal atypia; p = 0.001). All seven cases with gene amplification involved the lower limbs (six in the thigh and one in the lower leg). With regards to size, six of the seven cases with gene amplification occurred in tumours >100 mm in maximum dimension. There were no significant associations with patient age, gender, block:size ratio and radiological diagnosis. Of interest is that, although not reaching statistical significance (p = 0.227), there was no case of gene amplification in all 23 lesions arising within the subcutaneous fat. Multivariate logistic regression analysis was also performed to identify independent predictors of MDM2/ CDK4 amplification by introducing significant variables identified on univariate analysis as well as other co-variates of interest. Two regression models were analysed, the first considering only clinical and pathological factors (to account for institutions where imaging details are not readily available to the reporting pathologist) and the second considering clinical, radiological and pathological factors. Independent predictors of MDM2/CDK4 amplification in the first model were found to be histological features (presence of equivocal atypia; OR 12.5, 95%CI 1.9–80.3, p = 0.008) and macroscopic size (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and in the second model, histological features (presence of equivocal atypia; OR 16.7, 95%CI 2.0–139.2, p = 0.009) and tumour size determined on imaging (OR 4.1, 1.2–14.1, p = 0.026). Follow-up information was available in 107 (76.4%) patients, including surveillance imaging in 36 (33.6%), with a median follow-up time of 31 months (range 27.9–63.9 months). All seven amplified cases were included in this group with a median follow-up time of 24.7 months (range 3.1–50.5 months). There were four (10.8%) instances of local tumour recurrence (all occurring in MDM2/CDK4 nonamplified cases). In two (50%) of these cases, it was not possible to determine whether the lesions found on imaging represented residual disease or early recurrence. The mean and median macroscopic size of tumours which recurred was 86 mm and 93 mm, respectively (range 50–110 mm). All four cases were intramuscular in location. There were no distant metastases. Kaplan–Meier survival curves comparing the time to first local recurrence for the amplified and nonamplified groups are shown in Fig. 3. There was no statistically significant difference between the curves (p = 0.461).
DISCUSSION The utility of demonstrating MDM2/CDK4 gene amplification to support a morphological diagnosis of well differentiated liposarcoma is now well established based on the high sensitivity and specificity reported in multiple studies.6– 10,12,16 The purpose of this investigation, however, was to address the more common diagnostic problem of whether
206
Fig. 1
WONG et al.
Pathology (2016), 48(3), April
Correlation between tumour size as measured on imaging studies and on macroscopic examination.
Fig. 2 Lipomatous neoplasms included in this study were histologically consistent with a lipoma, comprising uniform mature adipocytes with occasional paucicellular
fibrous septa containing bland fibroblastic cells (A). Equivocal atypia was present in 12 (10.1%) cases in the form of isolated stromal cells with enlarged nuclei but without hyperchromasia (B). Cases with diagnostic atypia sufficient for the histological diagnosis of well differentiated liposarcoma were excluded from this study (C). MDM2 gene amplification was identified in 7 (5%) cases on interphase FISH analysis (D).
testing for these molecular changes is indicated in lipomatous neoplasms which, although histologically consistent with a lipoma, are concerning due to large size and/or a deep seated location. This study also aimed to identify clinical, radiological and pathological factors which might help predict for these alterations and avoid unnecessary costly and time consuming molecular testing. Although the use of
immunohistochemistry to detect MDM2/CDK4 protein overexpression as a surrogate marker of gene amplification has been previously proposed,3–5,11 in our experience and those of others,10 commercially available antibodies in routine practice are difficult to apply and reliably interpret. Brightfield dual colour in situ hybridisation to assess MDM2 amplification has also been recently introduced17 but is not in
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
Table 3 Case 1 2 3 4 5 6 7
207
Summary of clinicopathological characteristics of MDM2/CDK4 amplified cases Age, years
Sex
Site
Tissue plane
Size, mm
Imaging diagnosis
Microscopic findings
Amplification statusa
36 42 49 56 58 68 80
M M M M F F F
Left thigh Left thigh Left thigh Left thigh Left lower leg Right thigh Left thigh
IM Deep Deep Deep IM IM IM
190 50 160 105 125 140 150
LSb LSb LSb Lipoma Lipoma Lipoma NA
Minor atypia Chronic inflammation Simple lipoma Minor atypia Simple lipoma Minor atypia Simple lipoma
MDM2(×3), CDK4(×3) MDM2(×30), CDK4(×30) MDM2(×10), CDK4(×6) MDM2(×60) MDM2(×2), CDK4(×2) MDM2(×50) MDM2(×20)
IM, intramuscular; NA, not available. a b
Ratio of MDM2 and CDK4 to chromosome 12 centromere. LS liposarcoma favoured or cannot be confidently excluded.
Table 4
Results of univariate statistical analysis
Variable
Age, mean years Sex, % male Tumour site, % Upper limb Lower limb Trunk Head/neck Tissue plane, % subfascial Radiological size, mean mm Radiological diagnosis, % lipoma Macroscopic size, mean mm Block:size ratio Presence of minor atypia (microscopic) a
MDM2/CDK4 non-amplified 56.6 59.4% 35.3% 32.3% 30.8% 1.5% 82.6% 96.4 79.8% 93.1 1.2 6.8%
MDM2/CDK4 amplified 55.6 57.1% 0% 100% 0% 0% 100% 186.3 50% 131.4 1.1 42.9%
Significance
p = 0.851 p = 0.906 p = 0.004a
p = 0.227 p = 0.013a p = 0.088 p = 0.033a p = 0.879 p = 0.001a
Significant at 5% level.
common diagnostic use in most laboratories. Hence, like others,7–10,12,16,18 we believe that FISH currently remains the reference standard for assessment of MDM2/CDK4 status and was used as such in this study. Our findings based on a series of 140 cases show that MDM2/CDK4 amplification does occur in lipomatous neoplasms with histological features of a lipoma, although the incidence is low at 5% (95%CI 1.39–8.61%). Most cases were well sampled with the mean number of paraffin blocks submitted per centimetre of maximum tumour dimension (block:size ratio) being 1.2 ± 0.4. As lesional size is a major determinant in the assessment of adequacy of sampling, we compared the maximum tumour dimensions measured macroscopically to those measured on imaging studies and demonstrated excellent correlation (r = 0.74, 95%CI 0.64–0.82, p < 0.001), even in those specimens which were received piecemeal. Several factors were associated with MDM2/CDK4 gene amplification based on the results of univariate statistical analysis (Table 4). Tumour site was significant with all amplified cases involving the lower limbs (p = 0.004; six in the thigh and one in the lower leg). Tumour size measured both macroscopically and on imaging studies was also strongly associated (p = 0.033 and p = 0.013, respectively) with six of seven amplified cases measuring >100 mm in
maximum dimension. The presence equivocal atypia was the final significant association (p = 0.001). This atypia was seen in three of seven amplified cases and comprised isolated stromal cells with mild nuclear enlargement which we believe would comfortably fall within the acceptable spectrum of a simple lipoma, without showing the hyperchromasia diagnostic of well differentiated liposarcoma (Fig. 2). Of note is that amplified cases universally occurred in a subfascial location, although not reaching statistical significance (p = 0.227), with all 23 subcutaneous lesions lacking amplification. Importantly, there was no significant association with the block:size ratio, confirming that tissue sampling was not related to the finding of gene amplification. Multivariate logistic regression analysis was also performed to control for potential confounders and to identify independent predictors of MDM2/CDK4 amplification. Two models were analysed to account for the situation where the reporting pathologist may or may not have had imaging data readily available. In both models, the presence of equivocal atypia (OR 12.5, 95%CI 1.9–80.3, p = 0.008) and tumour size (as determined on imaging in the former model (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and as determined on macroscopic examination in the latter model (OR 4.1, 1.2–14.1, p = 0.026) remained significant covariates, after controlling for patient age, sex, tumour site, tissue plane involved and radiological diagnosis. The results suggest that apparently atypical imaging findings (such as thickened septa, abnormal lesional signal and contrast enhancement) lack specificity for MDM2/CDK4 amplification. This would also imply that having access to the imaging data adds little in this setting to assist the pathologist in selecting cases at greater risk of gene amplification, other than providing the radiological dimensions (which correlated with macroscopic size in this study). Our results confirm the findings of two recent studies which addressed similar but not identical issues.13,18 The first study, which examined an unselected series of histologically diagnosed lipomas and liposarcomas, recommended molecular testing for tumours with questionable cytological atypia, tumours >150 mm in size and relapsing lesions (not considered in our study).18 The second study, which validated guidelines developed at the authors’ institution for problematic lipomatous tumours, recommended FISH testing for deep seated lesions arising in the extremities measuring >100 mm in size in patients older than 50, those with equivocal atypia, recurrent lesions (not considered in our study) and for retroperitoneal, pelvic and abdominal tumours (not considered in our study).13
208
Fig. 3
WONG et al.
Pathology (2016), 48(3), April
Kaplan–Meier survival curves comparing time to first local recurrence in MDM2/CDK4 amplified and non-amplified cases. *Log-rank test.
A secondary objective in our study was to assess whether the finding of MDM2/CDK4 amplification in lesions which are histologically lipomas warrants separate classification as liposarcomas based on outcome data. This is an assumption which, in our experience, is widely held in practice and can also be seen in previously published investigations where ‘lipomas’ demonstrating gene amplification were reclassified as liposarcomas.6,13,19 This concept most likely represents an unsubstantiated extrapolation of early studies validating the finding of MDM2/CDK4 gene amplification in well differentiated liposarcoma, where almost always, the molecular alterations were accompanied by established atypical morphological features. Proponents of this school of thought often argue that it is the molecular changes that ultimately drive the biological behaviour of tumours and it is also well known that well differentiated liposarcomas can show large areas which histologically resemble lipomas, hence the socalled rubric of ‘lipoma-like liposarcoma’. Nevertheless, despite these theoretical considerations, there appears to be no or little empirical data, to our knowledge, to support the popular assumption that MDM2/CDK4 amplification occurring in a lipomatous lesion warrants its classification as a liposarcoma, even in the absence of morphological features befitting this diagnosis. In this study, follow-up information was available in 107 (76.4%) patients, including surveillance imaging in 36 (33.6%), with a median follow-up time of 31 months (range 27.9–63.9 months). There were four (10.8%) instances of local tumour recurrence, which interestingly all occurred in non-amplified cases. However, all four cases were intramuscular in location which is known to carry an increased risk of local recurrence.18 Kaplan–Meier survival curves for the amplified and non-amplified groups were not statistically different (p = 0.461, Fig. 3). Based on these findings,
therefore, there appears to be no definite evidence to indicate that MDM2/CDK4 amplified cases are at increased risk of early local tumour recurrence compared to those which are non-amplified. There are several limitations to this study which are openly acknowledged. First, the median follow-up time of 31 months is short for tumours which, although capable of early relapse,2 are known to recur up to many years following primary excision14 and it is likely that not all cases of tumour recurrence would have been adequately captured in this restricted time frame. Certainly, further studies with more prolonged follow-up times would be ideal to confirm the preliminary findings of the current investigation. In view of these uncertainties it would be prudent at present to continue to assess MDM2/CDK4 status in high risk cases and to recommend close clinical follow-up in those showing gene amplification until further evidence becomes available to help clarify these issues. Second, the small numbers of MDM2/ CDK4 amplified cases in this series prevented meaningful analysis of subgroups based on the level of gene amplification. It is probable that lesions with high level amplification are more likely to be biologically different to those with low level amplification, which at least in a proportion of cases, may represent a non-specific or spurious finding. It would be interesting in future studies to assess whether the former group demonstrates a clinical course more akin to bone fide liposarcomas. In summary, this study of 140 cases confirms that the incidence of MDM2/CDK4 amplification is low (5%) in large and/or deep-seated (subfascial) lipomatous neoplasms with histological features of a lipoma (or lipoma variant). Risk factors for gene amplification include tumour site (lower limbs, especially thigh), tumour size (>100 mm in maximum dimension either on macroscopic examination or imaging
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
studies) and the presence of equivocal atypia. All subcutaneous lesions were non-amplified. Only tumour size and the presence of equivocal atypia remained independent predictors of gene amplification on multivariate analysis and appreciation of these risk factors may help facilitate more selective FISH testing. Based on a mean follow-up period of 31 months, there was no statistically significant difference in the rate of local tumour recurrence between amplified and non-amplified tumours, although future studies with more prolonged follow-up would be required to confirm these preliminary findings. Conflicts of interest and sources of funding: This study was funded by a PathWest research grant. The authors state that there are no conflicts of interest to disclose. Address for correspondence: Daniel D. Wong, Department of Anatomical Pathology, PathWest Laboratory Medicine, QEII Medical Centre, Hospital Avenue, Nedlands, WA 6009, Australia. E-mail: [email protected]
References 1. Arrigoni G, Doglioni C. Atypical lipomatous tumor: molecular characterization. Curr Opin Oncol 2004; 16: 355–8. 2. Rosai J, Akerman M, Dal Cin P, et al. Combined morphologic and karyotypic study of 59 atypical lipomatous tumors. Evaluation of their relationship and differential diagnosis with other adipose tissue tumors (a report of the CHAMP Study Group). Am J Surg Pathol 1996; 20: 1182–9. 3. Aleixo PB, Hartmann AA, Menezes IC, et al. Can MDM2 and CDK4 make the diagnosis of well differentiated/dedifferentiated liposarcoma? An immunohistochemical study on 129 soft tissue tumours. J Clin Pathol 2009; 62: 1127–35. 4. Binh MB, Sastre-Garau X, Guillou L, et al. MDM2 and CDK4 immunostainings are useful adjuncts in diagnosing well-differentiated and dedifferentiated liposarcoma subtypes: a comparative analysis of 559 soft tissue neoplasms with genetic data. Am J Surg Pathol 2005; 29: 1340–7. 5. Dei Tos AP, Doglioni C, Piccinin S, et al. Coordinated expression and amplification of the MDM2, CDK4, and HMGI-C genes in atypical lipomatous tumours. J Pathol 2000; 190: 531–6. 6. Kashima T, Halai D, Ye H, et al. Sensitivity of MDM2 amplification and unexpected multiple faint alphoid 12 (alpha 12 satellite sequences) signals in atypical lipomatous tumor. Mod Pathol 2012; 25: 1384 –96.
209
7. Kimura H, Dobashi Y, Nojima T, et al. Utility of fluorescence in situ hybridization to detect MDM2 amplification in liposarcomas and their morphological mimics. Int J Clin Exp Pathol 2013; 6: 1306 – 16. 8. Putri RI, Siregar NC, Siregar B. Overexpression and amplification of Murine double minute 2 as a diagnostic tool in large lipomatous tumor and its correlation with Ki67 proliferation index: an institutional experience. Indian J Pathol Microbiol 2014; 57: 558–63. 9. Shimada S, Ishizawa T, Ishizawa K, et al. The value of MDM2 and CDK4 amplification levels using real-time polymerase chain reaction for the differential diagnosis of liposarcomas and their histologic mimickers. Hum Pathol 2006; 37: 1123–9. 10. Sirvent N, Coindre JM, Maire G, et al. Detection of MDM2-CDK4 amplification by fluorescence in situ hybridization in 200 paraffinembedded tumor samples: utility in diagnosing adipocytic lesions and comparison with immunohistochemistry and real-time PCR. Am J Surg Pathol 2007; 31: 1476–89. 11. Thway K, Flora R, Shah C, et al. Diagnostic utility of p16, CDK4, and MDM2 as an immunohistochemical panel in distinguishing welldifferentiated and dedifferentiated liposarcomas from other adipocytic tumors. Am J Surg Pathol 2012; 36: 462–9. 12. Weaver J, Downs-Kelly E, Goldblum JR, et al. Fluorescence in situ hybridization for MDM2 gene amplification as a diagnostic tool in lipomatous neoplasms. Mod Pathol 2008; 21: 943–9. 13. Clay MR, Martinez AP, Weiss SW, et al. MDM2 amplification in problematic lipomatous tumors: analysis of FISH testing criteria. Am J Surg Pathol 2015; 10: 7. 14. Weiss SW, Rao VK. Well-differentiated liposarcoma (atypical lipoma) of deep soft tissue of the extremities, retroperitoneum, and miscellaneous sites. A follow-up study of 92 cases with analysis of the incidence of “dedifferentiation”. Am J Surg Pathol 1992; 16: 1051 –8. 15. Ida CM, Wang X, Erickson-Johnson MR, et al. Primary retroperitoneal lipoma: a soft tissue pathology heresy?: report of a case with classic histologic, cytogenetics, and molecular genetic features. Am J Surg Pathol 2008; 32: 951–4. 16. Nishio J. Contributions of cytogenetics and molecular cytogenetics to the diagnosis of adipocytic tumors. J Biomed Biotechnol 2011; 2011: 524067. 17. Zhang W, McElhinny A, Nielsen A, et al. Automated brightfield dualcolor in situ hybridization for detection of mouse double minute 2 gene amplification in sarcomas. Appl Immunohistochem Mol Morphol 2011; 19: 54–61. 18. Zhang H, Erickson-Johnson M, Wang X, et al. Molecular testing for lipomatous tumors: critical analysis and test recommendations based on the analysis of 405 extremity-based tumors. Am J Surg Pathol 2010; 34: 1304–11. 19. Thway K, Wang J, Swansbury J, et al. Fluorescence in situ hybridization for MDM2 amplification as a routine ancillary diagnostic tool for suspected well-differentiated and dedifferentiated liposarcomas: experience at a tertiary center. Sarcoma 2015; 2015: 812089.
A N ATO M I C A L PAT H O L O G Y
MDM2/CDK4 gene amplification in large/deep-seated ‘lipomas’: incidence, predictors and clinical significance DANIEL D. WONG1,2, IRENE C. LOW1, JOANNE PEVERALL3, PETER D. ROBBINS1, DOMINIC V. SPAGNOLO1, ROBERT NAIRN4, RICHARD L. CAREY-SMITH5 AND DAVID WOOD5 1
Department of Anatomical Pathology, PathWest Laboratory Medicine, QEII Medical Centre, Nedlands, 2School of Pathology and Laboratory Medicine, University of Western Australia, Crawley, 3Department of Diagnostic Genomics, PathWest Laboratory Medicine, QEII Medical Centre, Nedlands, 4Department of Radiology, Sir Charles Gairdner Hospital, Nedlands, and 5Perth Orthopaedic Institute, Nedlands, WA, Australia
Summary This study of 140 cases assessed the incidence of MDM2/ CDK4 gene amplification in lipomatous neoplasms with histological features of a lipoma but which were of clinical concern due to large size (50 mm) and/or deep-seated (subfascial) location. Univariate and multivariate statistical analyses were used to identify clinical, radiological and pathological predictors of gene amplification. Differences in local recurrence rates between amplified and nonamplified cases were assessed using survival analysis. The findings indicate that the incidence of MDM2/CDK4 amplification in this setting is low at 5% (95%CI 1.4–8.6%). Variables associated with amplification on univariate analysis were tumour site (thigh, p = 0.004), size (>100 mm, p = 0.033) and presence of equivocal atypia (p = 0.001). Independent predictors on multivariate analysis were size (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and presence of equivocal atypia (OR 12.5, 95%CI 1.9–80.3, p = 0.008). There was no significant difference in local recurrence rates between amplified and non-amplified cases (p = 0.461) based on a median follow-up time of 31 months. Assessment for MDM2/CDK4 amplification, therefore, should be considered in ‘lipomas’ which are >100 mm in size, show equivocal atypia and arise in the thigh. However, the clinical significance of gene amplification in this setting is unclear and requires confirmation in larger studies. Key words: MDM2; CDK4; amplification; lipoma; liposarcoma. Received 16 October, revised 26 November, accepted 6 December 2015 Available online 2 March 2016
may be useful to help aid in this distinction. Specifically, well differentiated liposarcomas are cytogenetically characterised by giant marker and ring chromosomes involving the chromosomal region 12q13-15.1,2 MDM2 (murine double minute 2) and CDK4 (cyclin dependent kinase 4) are two of several genes known to be amplified in this region and detection of these genetic changes has demonstrated high sensitivity and specificity for the histological diagnosis of well differentiated liposarcoma.3–12 What is less certain, however, is whether testing for MDM2/CDK4 gene amplification is indicated in bland lipomatous neoplasms with histological features of a lipoma but which are concerning due to large size and/or a deepseated (subfascial) location. Whilst it would seem that this issue has been previously addressed, most studies which have examined these alterations in lipomatous neoplasms have only included small numbers of lipomas in their validation series.3–12 Equally uncertain is the clinical significance of demonstrating MDM2/CDK4 amplification in this setting. Although it is commonly assumed, extrapolating from early studies, that these molecular changes warrant a diagnosis of liposarcoma irrespective of the microscopic findings,6,13 there are presently few published empirical data to support this assumption. The aim of this study, therefore, was to address some of the issues related to this subject: first, to determine the incidence of MDM2/CDK4 amplification in a series of apparent ‘lipomas’ of large size and/or deep-seated (subfascial) location; second, to identify clinical, radiological and pathological factors which might predict for these alterations and help avoid unnecessary costly and time consuming molecular tests; and finally, to investigate whether the finding of gene amplification in this setting is associated with a more aggressive clinical course, warranting classification separate from simple lipomas.
INTRODUCTION
MATERIALS AND METHODS
Lipomas are among the most common soft tissue neoplasms encountered by surgical pathologists. Separating lipomas from well differentiated liposarcomas is usually not difficult in routine practice based on their microscopic appearances. In more problematic cases, assessment of genetic alterations
Well differentiated lipomatous neoplasms submitted for assessment of MDM2/CDK4 amplification by fluorescence in situ hybridisation (FISH) between January 2009 and August 2015 were retrospectively identified from our laboratory database. Inclusion criteria for this study were: (1) largest macroscopic dimension 50 mm and/or deep-seated (subfascial) location; (2) absence of diagnostic atypical morphological features which would lead to a
Print ISSN 0031-3025/Online ISSN 1465-3931 Crown Copyright © 2016 Published by Elsevier B.V. on behalf of The Royal College of Pathologists of Australasia. All rights reserved. DOI: http://dx.doi.org/10.1016/j.pathol.2016.02.007
204
WONG et al.
histological diagnosis of well-differentiated liposarcoma (namely, overtly atypical stromal cells showing enlarged and hyperchromatic nuclei with or without atypical lipoblasts). Retroperitoneal and pelvic lipomatous neoplasms, which have a marked predisposition for local recurrence even in the absence of atypical morphological features,14,15 were excluded. Cases with a prior diagnosis of liposarcoma and presenting with a recurrent mass were also excluded. Most tumours (n = 134, 95.7%) were reported by subspecialty soft tissue pathologists in our department (DDW, ICL, PDR and DVS). Haematoxylin and eosin (H&E) slides from all remaining cases were reviewed to confirm that the histological features satisfied the inclusion criteria for the study. Clinical and radiological data were obtained from pathology requisition forms and radiology reports. Variables considered included patient age, gender, tumour location, tissue plane involved (superficial, subfascial or intramuscular), tumour size (largest dimension) on imaging studies, imaging modality used [ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI)] and the presence of atypical imaging features (presence of thick septa, abnormal signal patterns and lesional enhancement following administration of contrast). The imaging diagnosis favoured by the reporting radiologist was noted, retrieved from the conclusion of radiology reports and in many cases, confirmed at our soft tissue multidisciplinary team meetings in which an experienced musculoskeletal radiologist was present. Follow-up information including time to first local recurrence as confirmed on surveillance imaging studies was obtained by interrogating the databases of radiological centres known to service our soft tissue surgeons who were involved in the management of the majority of the cases included in this study (n = 106, 75.7%). Pathological data considered included tumour size (largest dimension) on macroscopic examination, number of blocks submitted, the presence of unusual histological features (fat necrosis, inflammation, equivocal atypia in stromal cells falling short of that allowing a definite morphological diagnosis of liposarcoma) and the final histological diagnosis (including classification as a lipoma variant). For specimens which were received in multiple pieces, the maximum dimension of the largest piece was accepted as the best approximation for the overall lesion size (compared to aggregate dimensions which, in our experience, are less reproducible). To assess the adequacy of sampling, a block:size ratio was determined by dividing the number of blocks submitted for microscopic examination by the largest dimension of the tumour size in cm. Interphase FISH analysis to assess for the presence of MDM2/CDK4 amplification was performed on representative formalin fixed, paraffin embedded sections using in-house DNA probes specific for the MDM2 gene at 12q15, CDK4 gene at 12q15 and chromosome 12 centromere (CEP12). Areas of interest were targeted by correlating with serial H&E stained sections. Two hundred successive nuclei were examined by two independent analysts (including JP) using the Applied Imaging CytoVision image capture system (Leica Biosystems, Germany). MDM2:CEP12 and CDK4:CEP12 signal ratios 2 in more than 15% of tumour cells were considered amplified. Signal ratios between 2 and 10 were classified as low level amplification and ratios >10 as high level amplification. Statistical comparisons of proportions and means were performed using Pearson’s chi-square test and Student’s independent samples t-test, respectively. Pearson’s correlation was used to assess the strength of the correlation between the largest dimension of lesions measured macroscopically compared to those measured on imaging studies. Multivariate logistic regression analysis was then performed to identify any independent predictors of MDM2/CDK4 amplification. The forward stepwise conditional method was used to introduce selected variables of clinical interest and those found to be significant on univariate analysis (including age, gender, tumour site, tissue plane involved, tumour size, presence of equivocal atypia and final radiological diagnosis) into the multivariate equation. Kaplan–Meier survival curves were used to analyse the time to first local recurrence according to MDM2/CDK4 amplification status and the difference between the curves was examined using the log-rank test. Statistical analyses were performed using SPSS for Windows Version 20.0 (IBM Corporation, USA) with the two-sided statistical significance level set at 5%. This study was approved by the Institutional Review Board (QA# 9748). Informed consent was waived as there was no direct patient involvement and
Pathology (2016), 48(3), April
the results of this retrospective study were not anticipated to alter clinical management.
RESULTS A total of 140 cases were identified from our database meeting the inclusion criteria for the study, including 83 (59.3%) males and 57 (40.7%) females. The mean and median ages were 56.5 years and 57 years, respectively (range 21–88 years). Specific sites of the tumours are detailed in Table 1. The lower limbs were involved in 50 (35.7%) cases, upper limbs in 47 (33.6%), trunk (chest wall, abdominal wall and back) in 41 (29.3%) and head/neck region in two (1.4%). Of those involving the extremities, 65 (50.4%) involved the left side and 64 (49.6%) involved the right. Up to 23 (16.4%) were situated within the subcutaneous fat, 54 (38.6%) intramuscular and the remaining 62 (44.3%) were deep/subfascial but extramuscular. Imaging data were available for 100 (71.4%) cases, of which 82 (82%) were assessed with MRI, 10 (10%) with CT and 8 (8%) with ultrasound. Of these, the preferred radiological diagnosis by the reporting radiologist was that of a simple lipoma in 78 (78%), whereas a diagnosis of a welldifferentiated liposarcoma was either favoured or could not be confidently excluded in the remaining 22 (22%), based purely on the imaging characteristics, irrespective of size and location. The frequency of atypical imaging features (the presence of thick septa, abnormal signal and abnormal enhancement following contrast administration) in the latter group is detailed in Table 2. Up to 124 (88.6%) specimens were received intact. In the remaining 16 (11.4%) which were received in multiple pieces, tumour size was approximated by taking the maximum dimension of the largest piece. Mean and median tumour size on macroscopic examination was 95 mm and 85 mm, respectively (range 20–275 mm). Tumour dimensions as measured on imaging studies were available in 96 (68.6%) cases with a mean and median size of 102 mm and 90 mm, respectively (range 18–390 mm). Correlation between the macroscopic and radiologically determined dimensions for this subset was strong [r = 0.74, 95% confidence interval (CI) 0.64–0.82, p < 0.001; Fig. 1]. The average number of paraffin blocks submitted per centimetre of the maximum tumour dimension was 1.2 ± 0.4 (range 0.4–3.0). Microscopic examination of all 140 cases revealed features consistent with a simple lipoma in 119 (85%). Specifically, the lesions comprised solid sheets and large lobules of mature adipose tissue often forming circumscribed but encapsulated masses. Occasional thin, intervening, paucicellular fibrous septa containing isolated bland spindled fibroblastic cells were seen in most tumours (Fig. 2). Adipocytes displayed no more than mild variability in size and shape without the presence of well-developed univacuolated and multivacuolated lipoblasts. Fat necrosis featured in five (4.2%) cases whilst a patchy lymphocytic infiltrate was noted in two (1.7%). Ten (7.1%) showed more frequent and thicker paucicellular bands of fibrosis and were classified as fibrolipomas. Spindle cell lipomas accounted for eight (5.7%) cases showing the well-recognised features of a lipomatous neoplasm with plump, uniform CD34+ spindle cells set in a variably fibrous and myxoid stroma, with coarse collagen fibrils and mast cells. Other variants included osteolipomas [lipoma with metaplastic ossification, n = 2 (1.4%)] and a
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
Table 1
Clinicopathological characteristics of all included cases
Variable
Statistic
Mean age, years (range) (n = 140) Sex, male:female (n = 140) Tumour site (n = 140) Upper arm Lower arm/hand Thigh Lower leg/foot Trunk Head/neck Tissue plane (n = 139) Subcutaneous Intramuscular Subfascial (extramuscular) Macroscopic tumour size (n = 139) <50 mm 50–99 mm 100–150 mm >150 mm Histological classification (n = 140) Simple lipoma Fibrolipoma Spindle cell lipoma Other subtypes
56.5 (21–88) 3:2 35 12 42 8 41 2
(25%) (8.5%) (30%) (5.8%) (29.3%) (1.4%)
23 (16.5%) 54 (38.8%) 62 (44.6%) 16 70 35 18
(11.5%) (50.4%) (25.2%) (12.9%)
119 10 8 3
(85%) (7.1%) (5.7%) (2.1%)
hibernoma [conventional type, n = 1 (0.7%)]. By definition, none of the cases showed atypical stromal cells with enlarged and hyperchromatic nuclei, diagnostic of well differentiated liposarcoma. Of those classified as a simple lipoma, additional features of note included the presence of equivocal atypia (isolated stromal cells showing mild nuclear enlargement without hyperchromasia, falling short of the diagnostic features to allow a histological diagnosis of welldifferentiated liposarcoma, Fig. 2) in 12 (10.1%). MDM2/CDK4 amplification was assessed by FISH in all 140 cases. The MDM2 gene was amplified in seven (5%, 95%CI 1.4–8.6%) cases, with a median MDM2/CEP12 ratio of 20 (range 2–60). Three showed low level and four showed high level amplification, respectively. CDK4 was amplified in four (2.9%, 95%CI 0.1–5.7%) cases, with a median CDK4/CEP12 ratio of 4.5 (range 2–30). Three showed low
Table 2
Summary of imaging findings
Variable Imaging modality (n = 100) MRI CT US Radiological tumour size (n = 97) <50 mm 50–99 mm 100–150 mm >150 mm Atypical imaging features (n = 100) Thick septa Lesional enhancement Abnormal (non-fat) signal Favoured diagnosis (n = 100) Simple lipoma Liposarcomaa
Statistic
82 (82%) 10 (10%) 8 (8%) 7 47 26 17
(5%) (33.6%) (18.6%) (12.1%)
14 (14%) 12 (12%) 5 (5%) 78 (78%) 22 (22%)
CT, computed tomography; MRI, magnetic resonance imaging; US, ultrasound. a
Liposarcoma favoured or could not be confidently excluded.
205
level and one showed high level amplification, respectively. MDM2 and CDK4 were co-amplified in four cases. The clinicopathological characteristics of amplified cases are summarised in Table 3. Univariate statistical analysis was performed to investigate the association of various clinical, radiological and pathological factors with MDM2/CDK4 amplification. The results of this analysis are detailed in Table 4. In summary, factors significantly associated with gene amplification were tumour site (p = 0.004), tumour size (both macroscopic and on imaging studies; p = 0.033 and p = 0.013, respectively) and histological features (presence of equivocal atypia; p = 0.001). All seven cases with gene amplification involved the lower limbs (six in the thigh and one in the lower leg). With regards to size, six of the seven cases with gene amplification occurred in tumours >100 mm in maximum dimension. There were no significant associations with patient age, gender, block:size ratio and radiological diagnosis. Of interest is that, although not reaching statistical significance (p = 0.227), there was no case of gene amplification in all 23 lesions arising within the subcutaneous fat. Multivariate logistic regression analysis was also performed to identify independent predictors of MDM2/ CDK4 amplification by introducing significant variables identified on univariate analysis as well as other co-variates of interest. Two regression models were analysed, the first considering only clinical and pathological factors (to account for institutions where imaging details are not readily available to the reporting pathologist) and the second considering clinical, radiological and pathological factors. Independent predictors of MDM2/CDK4 amplification in the first model were found to be histological features (presence of equivocal atypia; OR 12.5, 95%CI 1.9–80.3, p = 0.008) and macroscopic size (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and in the second model, histological features (presence of equivocal atypia; OR 16.7, 95%CI 2.0–139.2, p = 0.009) and tumour size determined on imaging (OR 4.1, 1.2–14.1, p = 0.026). Follow-up information was available in 107 (76.4%) patients, including surveillance imaging in 36 (33.6%), with a median follow-up time of 31 months (range 27.9–63.9 months). All seven amplified cases were included in this group with a median follow-up time of 24.7 months (range 3.1–50.5 months). There were four (10.8%) instances of local tumour recurrence (all occurring in MDM2/CDK4 nonamplified cases). In two (50%) of these cases, it was not possible to determine whether the lesions found on imaging represented residual disease or early recurrence. The mean and median macroscopic size of tumours which recurred was 86 mm and 93 mm, respectively (range 50–110 mm). All four cases were intramuscular in location. There were no distant metastases. Kaplan–Meier survival curves comparing the time to first local recurrence for the amplified and nonamplified groups are shown in Fig. 3. There was no statistically significant difference between the curves (p = 0.461).
DISCUSSION The utility of demonstrating MDM2/CDK4 gene amplification to support a morphological diagnosis of well differentiated liposarcoma is now well established based on the high sensitivity and specificity reported in multiple studies.6– 10,12,16 The purpose of this investigation, however, was to address the more common diagnostic problem of whether
206
Fig. 1
WONG et al.
Pathology (2016), 48(3), April
Correlation between tumour size as measured on imaging studies and on macroscopic examination.
Fig. 2 Lipomatous neoplasms included in this study were histologically consistent with a lipoma, comprising uniform mature adipocytes with occasional paucicellular
fibrous septa containing bland fibroblastic cells (A). Equivocal atypia was present in 12 (10.1%) cases in the form of isolated stromal cells with enlarged nuclei but without hyperchromasia (B). Cases with diagnostic atypia sufficient for the histological diagnosis of well differentiated liposarcoma were excluded from this study (C). MDM2 gene amplification was identified in 7 (5%) cases on interphase FISH analysis (D).
testing for these molecular changes is indicated in lipomatous neoplasms which, although histologically consistent with a lipoma, are concerning due to large size and/or a deep seated location. This study also aimed to identify clinical, radiological and pathological factors which might help predict for these alterations and avoid unnecessary costly and time consuming molecular testing. Although the use of
immunohistochemistry to detect MDM2/CDK4 protein overexpression as a surrogate marker of gene amplification has been previously proposed,3–5,11 in our experience and those of others,10 commercially available antibodies in routine practice are difficult to apply and reliably interpret. Brightfield dual colour in situ hybridisation to assess MDM2 amplification has also been recently introduced17 but is not in
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
Table 3 Case 1 2 3 4 5 6 7
207
Summary of clinicopathological characteristics of MDM2/CDK4 amplified cases Age, years
Sex
Site
Tissue plane
Size, mm
Imaging diagnosis
Microscopic findings
Amplification statusa
36 42 49 56 58 68 80
M M M M F F F
Left thigh Left thigh Left thigh Left thigh Left lower leg Right thigh Left thigh
IM Deep Deep Deep IM IM IM
190 50 160 105 125 140 150
LSb LSb LSb Lipoma Lipoma Lipoma NA
Minor atypia Chronic inflammation Simple lipoma Minor atypia Simple lipoma Minor atypia Simple lipoma
MDM2(×3), CDK4(×3) MDM2(×30), CDK4(×30) MDM2(×10), CDK4(×6) MDM2(×60) MDM2(×2), CDK4(×2) MDM2(×50) MDM2(×20)
IM, intramuscular; NA, not available. a b
Ratio of MDM2 and CDK4 to chromosome 12 centromere. LS liposarcoma favoured or cannot be confidently excluded.
Table 4
Results of univariate statistical analysis
Variable
Age, mean years Sex, % male Tumour site, % Upper limb Lower limb Trunk Head/neck Tissue plane, % subfascial Radiological size, mean mm Radiological diagnosis, % lipoma Macroscopic size, mean mm Block:size ratio Presence of minor atypia (microscopic) a
MDM2/CDK4 non-amplified 56.6 59.4% 35.3% 32.3% 30.8% 1.5% 82.6% 96.4 79.8% 93.1 1.2 6.8%
MDM2/CDK4 amplified 55.6 57.1% 0% 100% 0% 0% 100% 186.3 50% 131.4 1.1 42.9%
Significance
p = 0.851 p = 0.906 p = 0.004a
p = 0.227 p = 0.013a p = 0.088 p = 0.033a p = 0.879 p = 0.001a
Significant at 5% level.
common diagnostic use in most laboratories. Hence, like others,7–10,12,16,18 we believe that FISH currently remains the reference standard for assessment of MDM2/CDK4 status and was used as such in this study. Our findings based on a series of 140 cases show that MDM2/CDK4 amplification does occur in lipomatous neoplasms with histological features of a lipoma, although the incidence is low at 5% (95%CI 1.39–8.61%). Most cases were well sampled with the mean number of paraffin blocks submitted per centimetre of maximum tumour dimension (block:size ratio) being 1.2 ± 0.4. As lesional size is a major determinant in the assessment of adequacy of sampling, we compared the maximum tumour dimensions measured macroscopically to those measured on imaging studies and demonstrated excellent correlation (r = 0.74, 95%CI 0.64–0.82, p < 0.001), even in those specimens which were received piecemeal. Several factors were associated with MDM2/CDK4 gene amplification based on the results of univariate statistical analysis (Table 4). Tumour site was significant with all amplified cases involving the lower limbs (p = 0.004; six in the thigh and one in the lower leg). Tumour size measured both macroscopically and on imaging studies was also strongly associated (p = 0.033 and p = 0.013, respectively) with six of seven amplified cases measuring >100 mm in
maximum dimension. The presence equivocal atypia was the final significant association (p = 0.001). This atypia was seen in three of seven amplified cases and comprised isolated stromal cells with mild nuclear enlargement which we believe would comfortably fall within the acceptable spectrum of a simple lipoma, without showing the hyperchromasia diagnostic of well differentiated liposarcoma (Fig. 2). Of note is that amplified cases universally occurred in a subfascial location, although not reaching statistical significance (p = 0.227), with all 23 subcutaneous lesions lacking amplification. Importantly, there was no significant association with the block:size ratio, confirming that tissue sampling was not related to the finding of gene amplification. Multivariate logistic regression analysis was also performed to control for potential confounders and to identify independent predictors of MDM2/CDK4 amplification. Two models were analysed to account for the situation where the reporting pathologist may or may not have had imaging data readily available. In both models, the presence of equivocal atypia (OR 12.5, 95%CI 1.9–80.3, p = 0.008) and tumour size (as determined on imaging in the former model (OR 3.9, 95%CI 1.4–11.3, p = 0.012) and as determined on macroscopic examination in the latter model (OR 4.1, 1.2–14.1, p = 0.026) remained significant covariates, after controlling for patient age, sex, tumour site, tissue plane involved and radiological diagnosis. The results suggest that apparently atypical imaging findings (such as thickened septa, abnormal lesional signal and contrast enhancement) lack specificity for MDM2/CDK4 amplification. This would also imply that having access to the imaging data adds little in this setting to assist the pathologist in selecting cases at greater risk of gene amplification, other than providing the radiological dimensions (which correlated with macroscopic size in this study). Our results confirm the findings of two recent studies which addressed similar but not identical issues.13,18 The first study, which examined an unselected series of histologically diagnosed lipomas and liposarcomas, recommended molecular testing for tumours with questionable cytological atypia, tumours >150 mm in size and relapsing lesions (not considered in our study).18 The second study, which validated guidelines developed at the authors’ institution for problematic lipomatous tumours, recommended FISH testing for deep seated lesions arising in the extremities measuring >100 mm in size in patients older than 50, those with equivocal atypia, recurrent lesions (not considered in our study) and for retroperitoneal, pelvic and abdominal tumours (not considered in our study).13
208
Fig. 3
WONG et al.
Pathology (2016), 48(3), April
Kaplan–Meier survival curves comparing time to first local recurrence in MDM2/CDK4 amplified and non-amplified cases. *Log-rank test.
A secondary objective in our study was to assess whether the finding of MDM2/CDK4 amplification in lesions which are histologically lipomas warrants separate classification as liposarcomas based on outcome data. This is an assumption which, in our experience, is widely held in practice and can also be seen in previously published investigations where ‘lipomas’ demonstrating gene amplification were reclassified as liposarcomas.6,13,19 This concept most likely represents an unsubstantiated extrapolation of early studies validating the finding of MDM2/CDK4 gene amplification in well differentiated liposarcoma, where almost always, the molecular alterations were accompanied by established atypical morphological features. Proponents of this school of thought often argue that it is the molecular changes that ultimately drive the biological behaviour of tumours and it is also well known that well differentiated liposarcomas can show large areas which histologically resemble lipomas, hence the socalled rubric of ‘lipoma-like liposarcoma’. Nevertheless, despite these theoretical considerations, there appears to be no or little empirical data, to our knowledge, to support the popular assumption that MDM2/CDK4 amplification occurring in a lipomatous lesion warrants its classification as a liposarcoma, even in the absence of morphological features befitting this diagnosis. In this study, follow-up information was available in 107 (76.4%) patients, including surveillance imaging in 36 (33.6%), with a median follow-up time of 31 months (range 27.9–63.9 months). There were four (10.8%) instances of local tumour recurrence, which interestingly all occurred in non-amplified cases. However, all four cases were intramuscular in location which is known to carry an increased risk of local recurrence.18 Kaplan–Meier survival curves for the amplified and non-amplified groups were not statistically different (p = 0.461, Fig. 3). Based on these findings,
therefore, there appears to be no definite evidence to indicate that MDM2/CDK4 amplified cases are at increased risk of early local tumour recurrence compared to those which are non-amplified. There are several limitations to this study which are openly acknowledged. First, the median follow-up time of 31 months is short for tumours which, although capable of early relapse,2 are known to recur up to many years following primary excision14 and it is likely that not all cases of tumour recurrence would have been adequately captured in this restricted time frame. Certainly, further studies with more prolonged follow-up times would be ideal to confirm the preliminary findings of the current investigation. In view of these uncertainties it would be prudent at present to continue to assess MDM2/CDK4 status in high risk cases and to recommend close clinical follow-up in those showing gene amplification until further evidence becomes available to help clarify these issues. Second, the small numbers of MDM2/ CDK4 amplified cases in this series prevented meaningful analysis of subgroups based on the level of gene amplification. It is probable that lesions with high level amplification are more likely to be biologically different to those with low level amplification, which at least in a proportion of cases, may represent a non-specific or spurious finding. It would be interesting in future studies to assess whether the former group demonstrates a clinical course more akin to bone fide liposarcomas. In summary, this study of 140 cases confirms that the incidence of MDM2/CDK4 amplification is low (5%) in large and/or deep-seated (subfascial) lipomatous neoplasms with histological features of a lipoma (or lipoma variant). Risk factors for gene amplification include tumour site (lower limbs, especially thigh), tumour size (>100 mm in maximum dimension either on macroscopic examination or imaging
MDM2/CDK4 AMPLIFICATION IN LARGE/DEEP LIPOMAS
studies) and the presence of equivocal atypia. All subcutaneous lesions were non-amplified. Only tumour size and the presence of equivocal atypia remained independent predictors of gene amplification on multivariate analysis and appreciation of these risk factors may help facilitate more selective FISH testing. Based on a mean follow-up period of 31 months, there was no statistically significant difference in the rate of local tumour recurrence between amplified and non-amplified tumours, although future studies with more prolonged follow-up would be required to confirm these preliminary findings. Conflicts of interest and sources of funding: This study was funded by a PathWest research grant. The authors state that there are no conflicts of interest to disclose. Address for correspondence: Daniel D. Wong, Department of Anatomical Pathology, PathWest Laboratory Medicine, QEII Medical Centre, Hospital Avenue, Nedlands, WA 6009, Australia. E-mail: [email protected]
References 1. Arrigoni G, Doglioni C. Atypical lipomatous tumor: molecular characterization. Curr Opin Oncol 2004; 16: 355–8. 2. Rosai J, Akerman M, Dal Cin P, et al. Combined morphologic and karyotypic study of 59 atypical lipomatous tumors. Evaluation of their relationship and differential diagnosis with other adipose tissue tumors (a report of the CHAMP Study Group). Am J Surg Pathol 1996; 20: 1182–9. 3. Aleixo PB, Hartmann AA, Menezes IC, et al. Can MDM2 and CDK4 make the diagnosis of well differentiated/dedifferentiated liposarcoma? An immunohistochemical study on 129 soft tissue tumours. J Clin Pathol 2009; 62: 1127–35. 4. Binh MB, Sastre-Garau X, Guillou L, et al. MDM2 and CDK4 immunostainings are useful adjuncts in diagnosing well-differentiated and dedifferentiated liposarcoma subtypes: a comparative analysis of 559 soft tissue neoplasms with genetic data. Am J Surg Pathol 2005; 29: 1340–7. 5. Dei Tos AP, Doglioni C, Piccinin S, et al. Coordinated expression and amplification of the MDM2, CDK4, and HMGI-C genes in atypical lipomatous tumours. J Pathol 2000; 190: 531–6. 6. Kashima T, Halai D, Ye H, et al. Sensitivity of MDM2 amplification and unexpected multiple faint alphoid 12 (alpha 12 satellite sequences) signals in atypical lipomatous tumor. Mod Pathol 2012; 25: 1384 –96.
209
7. Kimura H, Dobashi Y, Nojima T, et al. Utility of fluorescence in situ hybridization to detect MDM2 amplification in liposarcomas and their morphological mimics. Int J Clin Exp Pathol 2013; 6: 1306 – 16. 8. Putri RI, Siregar NC, Siregar B. Overexpression and amplification of Murine double minute 2 as a diagnostic tool in large lipomatous tumor and its correlation with Ki67 proliferation index: an institutional experience. Indian J Pathol Microbiol 2014; 57: 558–63. 9. Shimada S, Ishizawa T, Ishizawa K, et al. The value of MDM2 and CDK4 amplification levels using real-time polymerase chain reaction for the differential diagnosis of liposarcomas and their histologic mimickers. Hum Pathol 2006; 37: 1123–9. 10. Sirvent N, Coindre JM, Maire G, et al. Detection of MDM2-CDK4 amplification by fluorescence in situ hybridization in 200 paraffinembedded tumor samples: utility in diagnosing adipocytic lesions and comparison with immunohistochemistry and real-time PCR. Am J Surg Pathol 2007; 31: 1476–89. 11. Thway K, Flora R, Shah C, et al. Diagnostic utility of p16, CDK4, and MDM2 as an immunohistochemical panel in distinguishing welldifferentiated and dedifferentiated liposarcomas from other adipocytic tumors. Am J Surg Pathol 2012; 36: 462–9. 12. Weaver J, Downs-Kelly E, Goldblum JR, et al. Fluorescence in situ hybridization for MDM2 gene amplification as a diagnostic tool in lipomatous neoplasms. Mod Pathol 2008; 21: 943–9. 13. Clay MR, Martinez AP, Weiss SW, et al. MDM2 amplification in problematic lipomatous tumors: analysis of FISH testing criteria. Am J Surg Pathol 2015; 10: 7. 14. Weiss SW, Rao VK. Well-differentiated liposarcoma (atypical lipoma) of deep soft tissue of the extremities, retroperitoneum, and miscellaneous sites. A follow-up study of 92 cases with analysis of the incidence of “dedifferentiation”. Am J Surg Pathol 1992; 16: 1051 –8. 15. Ida CM, Wang X, Erickson-Johnson MR, et al. Primary retroperitoneal lipoma: a soft tissue pathology heresy?: report of a case with classic histologic, cytogenetics, and molecular genetic features. Am J Surg Pathol 2008; 32: 951–4. 16. Nishio J. Contributions of cytogenetics and molecular cytogenetics to the diagnosis of adipocytic tumors. J Biomed Biotechnol 2011; 2011: 524067. 17. Zhang W, McElhinny A, Nielsen A, et al. Automated brightfield dualcolor in situ hybridization for detection of mouse double minute 2 gene amplification in sarcomas. Appl Immunohistochem Mol Morphol 2011; 19: 54–61. 18. Zhang H, Erickson-Johnson M, Wang X, et al. Molecular testing for lipomatous tumors: critical analysis and test recommendations based on the analysis of 405 extremity-based tumors. Am J Surg Pathol 2010; 34: 1304–11. 19. Thway K, Wang J, Swansbury J, et al. Fluorescence in situ hybridization for MDM2 amplification as a routine ancillary diagnostic tool for suspected well-differentiated and dedifferentiated liposarcomas: experience at a tertiary center. Sarcoma 2015; 2015: 812089.