PET to assess early metabolic response and to guide treatment of adenocarcinoma of the oesophagogastric junction: the MUNICON phase II trial

Articles

PET to assess early metabolic response and to guide treatment of adenocarcinoma of the oesophagogastric junction: the MUNICON phase II trial Florian Lordick, Katja Ott, Bernd-Joachim Krause, Wolfgang A Weber, Karen Becker, Hubert J Stein, Sylvie Lorenzen, Tibor Schuster, Hinrich Wieder, Ken Herrmann, Rainer Bredenkamp, Heinz Höfler, Ulrich Fink, Christian Peschel, Markus Schwaiger, Jörg R Siewert

Summary Background In patients with locally advanced adenocarcinoma of the oesophagogastric junction (AEG), early metabolic response defined by 18-fluorodeoxyglucose-PET ([18F]FDG-PET) during neoadjuvant chemotherapy is predictive of histopathological response and survival. We aimed to assess the feasibility of a PET-response-guided treatment algorithm and its potential effect on prognosis. Methods Between May 27, 2002, and Aug 4, 2005, 119 patients with locally advanced adenocarcinoma of AEG type 1 (distal oesophageal adenocarcinoma) or type 2 (gastric cardia adenocarcinoma) were recruited into this prospective, single-centre study. All patients were assigned to 2 weeks of platinum and fluorouracil-based induction chemotherapy (evaluation period). Those with decreases in tumour glucose standard uptake values (SUVs), predefined as decreases of 35% or more at the end of the evaluation period and measured by PET, were defined as metabolic responders. Responders continued to receive neoadjuvant chemotherapy of folinic acid and fluorouracil plus cisplatin, or folinic acid and fluorouracil plus cisplatin and paclitaxel, or folinic acid and fluorouracil plus oxaliplatin for 12 weeks and then proceeded to surgery. Metabolic non-responders discontinued chemotherapy after the 2-week evaluation period and proceeded to surgery. The primary endpoint was median overall survival of metabolic responders and nonresponders. Secondary endpoints were median event-free survival, postoperative complications and mortality, number of residual tumour-free (R0) resections, and histopathological responses. This study has been registered in the European Clinical Trials Database (EudraCT) as trial 2007-003356-11. Findings 110 patients were evaluable for metabolic responses. 54 of these patients had metabolic responses (ie, decrease of 35% or more in tumour glucose SUV) after 2 weeks of induction chemotherapy, corresponding to a response of 49% (95% CI 39–59). 104 patients had tumour resection (50 in the responder group and 54 in the nonresponder group). After a median follow-up of 2·3 years (IQR 1·7–3·0), median overall survival was not reached in metabolic responders, whereas median overall survival was 25·8 months (19·4–32·2) in non-responders (HR 2·13 [1·14–3·99, p=0·015). Median event-free survival was 29·7 months (95% CI 23·6–35·7) in metabolic responders and 14·1 months (7·5–20·6) in non-responders (hazard ratio [HR] 2·18 [1·32–3·62], p=0·002). Major histological remissions (<10% residual tumour) were noted in 29 of 50 metabolic responders (58% [95% CI 48–67]), but no histological response was noted in metabolic non-responders. Interpretation This study confirmed prospectively the usefulness of early metabolic response evaluation, and shows the feasibility of a PET-guided treatment algorithm. These findings might enable tailoring of multimodal treatment in accordance with individual tumour biology in future randomised trials.

Introduction After the publication of two randomised controlled trials,1–3 neoadjuvant chemotherapy has become an accepted choice for the treatment of locally advanced adenocarcinoma of the oesophagus and the oesophagogastric junction. However, for patients who do not respond, the prognosis after neoadjuvant chemotherapy might be worse than that of a primarily surgical approach.4 Additionally, inefficient neoadjuvant treatment leads to adverse events, allows tumour progression during chemotherapy, costs time, and increases health expenses. Therefore, the ability to predict response to chemotherapy is clearly desirable. Measurement of early changes in tumour glucose uptake by use of 18-fluorodeoxyglucose-PET ([18F]FDGhttp://oncology.thelancet.com Vol 8 September 2007

PET) has yielded reproducible results that are useful for predicting clinical and histopathological response after multiple-course neoadjuvant chemotherapy.5,6 Specifically, we have noted that when a quantitative threshold for metabolic response is used, PET identifies accurately non-responding tumours within 2 weeks of treatment initiation. This finding suggests that PET can be used to tailor treatment to individual patients. Studies that show the use of PET results for modifying treatment are currently absent, despite the promising findings of studies evaluating PET for monitoring tumour response during treatment.7 By contrast to previous studies from our institution that assessed and validated the optimum cut-off values for the early prediction of response,5,6 the study presented here (the Metabolic

Lancet Oncol 2007: 8: 797–805 Published Online August 10, 2007 DOI:10.1016/S14702045(07)70244-9 See Reflection and Reaction page 754 Department of Surgery (F Lordick MD, K Ott MD, Prof U Fink MD, Prof J R Siewert MD), Nuclear Medicine (Prof B-J Krause MD, W A Weber MD, H Wieder MD, K Herrmann MD, Prof M Schwaiger MD), Third Department of Medicine— Haematology and Medical Oncology (F Lordick, S Lorenzen MD, Prof C Peschel MD), Department of Pathology (K Becker MD, Prof H Höfler MD), Department of Medical Statistics (T Schuster MS), and Munich Center for Clinical Studies (R Bredenkamp BSc), Clinic rechts der Isar, Technical University of Munich, Munich, Germany; and Department of Surgery, Paracelsus Private Medical University, Salzburg, Austria (Prof H J Stein MD) Correspondence to: Dr Florian Lordick, National Centre for Tumour Diseases, University of Heidelberg, Im Neuenheimer Feld 350, D 69120 Heidelberg, Germany fl[email protected]

797

Articles

PET day 0

Chemotherapy

PET day 14 Responder

Chemotherapy for 12 weeks

Non-responder

Resection

The study was assigned the number 2007-003356-11 in the European Clinical Trials Database (EudraCT). Patients underwent a baseline PET assessment within 1 week preceding the start of neoadjuvant chemotherapy. Patients were eligible for inclusion only if PET scans showed sufficient contrast between tumour and surrounding tissues; this was based on standard uptake value (SUV) measurements: SUVtumour>1·35×SUVliver+2× standard deviation SUVliver. SUV is a quantitative measure of the intracellular glucose uptake. Clinically, SUV is used to image the viability of tumour tissue. The PET was repeated 14 days after the start of chemotherapy. Patients whose tumour SUV had decreased by 35% or more, which indicated a clear decrease in tumour glucose uptake and viability, were defined as metabolic responders.5,6 This cut-off was predefined in the study protocol and based on the authors’ previous research. These patients continued to receive chemotherapy for a maximum duration of 12 weeks before undergoing surgery. Metabolic non-responders discontinued chemotherapy and proceeded to surgery (figure 1).

PET Resection

Figure 1: Study design

response evalUatioN for Individualisation of neoadjuvant Chemotherapy in oesOphageal and oesophagogastric adeNocarcinoma (MUNICON) trial prospectively evaluated the feasibility and potential effect on prognosis of administering PET-response-guided chemotherapy to patients with locally advanced adenocarcinoma of the oesophagus and the oesophagogastric junction.

Methods Patients Patients with locally advanced adenocarcinoma of the oesophagogastric junction (AEG) type 1 (distal oesophageal adenocarcinoma) or type 2 (gastric cardia adenocarcinoma) according to Siewert’s classification8) were eligible. Patients were staged as cT3 or cT4 based on CT and endoscopic ultrasonography. Haematogenous metastases were excluded by PET. Exclusion criteria were: medical contraindications against chemotherapy with platinum plus fluorouracil; or unacceptable risks for oesophagectomy, as indicated by Bartel’s score.9 All patients were discussed by a multidisciplinary tumour board. None of the patients had been included into any of the previously published studies.5,6 All participants were aged 18 years or older and provided written informed consent.

Procedures This trial was a single-centre, exploratory phase II study. The protocol was approved by the local ethics committee. 798

A baseline PET scan with an ECAT EXACT full-ring PET scanner (CTI-Siemens, Knoxville, TN, USA) was done on all but four study participants as part of the staging procedure. In four patients, PET scanning was done with a Sensation 16 Biograph PET/CT scanner (CTI-Siemens, Knoxville, TN, USA); however, for all patients, the same machine was used for their baseline and 2-week scans. Patients fasted for 6 h before the PET scan. Blood glucose concentrations were measured before each PET scan, and all values measured less than 8·33 mmol/L. Static emission images of the tumour region (20-min duration) were acquired 40 min after intravenous injection with 300–400 MBq of [18F]FDG. All PET scans were acquired in 3-dimensional mode with an acquisition time of 3 min for each bed position. Image data were corrected for dead time and random events by iterative reconstruction (four iterations, eight subsets) by use of attenuation-weighted, ordered subset, and maximisation algorithms. Attenuation-corrected emission data were reconstructed by filtered back-projection by use of a Hanning filter with a cut-off frequency of 0·4 cycles per pixel. Image data were normalised for injected dose of [18F]FDG and patients’ body-surface area, resulting in parametric images representing regional SUVs. The spatial resolution of the reconstructed images was 6–8 mm at full width, half maximum. For quantitative measurement, a circular region of interest (diameter 1·5 cm, corresponding to 10 pixels) was placed over the tumour in the slice with maximum [18F]FDG uptake in the baseline scan. In the second PET scan, the region of interest was placed at the same position as in the baseline study by use of the anatomical landmarks of the transmission image as a reference. The decline in SUV was calculated by comparison of the baseline SUV and http://oncology.thelancet.com Vol 8 September 2007

Articles

the SUV measurements taken at 2 weeks. Data reconstruction and measurement of the SUVs were done as described previously.5,6

pathologists and was specified according to published criteria.11

Follow-up Chemotherapy Neoadjuvant chemotherapy consisted of two cycles of cisplatin (Medac GmbH, Wedel, Germany; 50 mg/m²) given on days 1, 15, and 29 (1 h infusion time), and folinic acid (Medac GmbH, Wedel, Germany; 500 mg/m² over 2 h) plus fluorouracil (Hexal AG, Holzkirchen, Germany; 2000 mg/m² over 24 h) on days 1, 8, 15, 22, 29, and 36, all repeated on day 49. For patients with a glomerular filtration rate of less than 60 mL/kg/min, oxaliplatin (Sanofi-Aventis Group, Paris, France; 85 mg/m² over 2 h) replaced cisplatin. Patients aged 60 years or younger with a good health status were additionally given paclitaxel (Bristol-Myers Squibb, New York, NY, USA; 80 mg/m² over 3 h) on days 0, 14, and 28. Endoscopy and CT were repeated before and after the second cycle of chemotherapy to exclude tumour progression. Adverse events were reported according to the National Cancer Institute Criteria, version 3.0.

Surgery Patients with AEG type 1 tumours underwent abdominothoracic oesophagectomy. Those with AEG type 2 tumours had transhiatal extended gastrectomy if the resection margins were tumour-free, or underwent additional abdominothoracic oesophagectomy if the resection margins contained tumour tissue. Surgery was planned to be undertaken within 2 weeks after the last administration of chemotherapy in non-responders and within 4 weeks after the last administration of chemotherapy in responders. A longer recovery period before surgery was scheduled for responding patients to enable adequate recovery of organ functions after 12 weeks of chemotherapy, compared with non-responding patients who had received only 2 weeks of chemotherapy. Additionally, the study protocol required that all staging investigations, including upper gastrointestinal endoscopy and CT, but not PET, were repeated preoperatively in responding patients.

Patient follow-up included CT and endoscopy done at 3-month intervals during the first year after surgery and at 6-month intervals thereafter. Survival was calculated from the day of study inclusion. Event-free survival was Responders (n=54)

Non-responders (n=56)

Median age, years (IQR)

62 (53–69)

61 (52–69)

0·684

Men, n

51

52

0·733

ECOG-PS 0, n

50

46

0·152

Localisation, AEG type 1, n

38

37

0·685

Intestinal type according to Laurén

46

46

0·798

T3 category, n

54

56

1·000

N0 category, n

10

6

0·135

Grading G3, n

34

25

Median SUV (tumour) (IQR)

8·3 (6·3–11·0)

p

0·059

6·8 (5·1–9·0)

0·018

ECOG-PS=Eastern Cooperative Oncology Group-Performance Status. AEG=adenocarcinoma of the oesophagogastric junction. SUV=standard uptake value.

Table 1: Baseline characteristics of patients assessed with PET for metabolic response to neoadjuvant chemotherapy

n

Metabolic response, n (%)

95% CI

Cisplatin, folinic acid, and fluorouracil

48

26 (54)

39–69

Oxaliplatin, folinic acid, and fluorouracil

17

8 (47)

23–72

Paclitaxel, cisplatin, folinic acid, and fluorouracil

45

20 (44)

30–60

Table 2: Metabolic responses with different chemotherapy regimens in 110 evaluable patients

Responder (n=50)

Non-responder (n=54)

p

R0

48 (96)

40 (74)

0·002

R1

2 (4)

14 (26)

..

Resection margin, n (%)

Histopathological response*, n (%) Score 1 (a+b)

29 (58)

0

Score 2

10 (20)

2 (4)

0·001 ..

Score 3

11 (22)

52 (96)

..

Pathology

pT category, n (%)

Tumour regression was assessed by two pathologists (KB or HH) by use of a recently published histopathological scoring system:10 patients with less than 10% residual tumour (score 1a: 0% residual tumour, complete remission; score 1b: <10% residual tumour, subtotal remission) were classified as responders. All other patients were classified as non-responders, including score 2 (10–50% residual tumour) and score 3 (>50% residual tumour). All specimens were seen by two pathologists. Discrepancies in histological response classification or in any other substantial findings were discussed and histology slides were reviewed until an agreement was reached. The involvement of the oral, aboral, and circumferential resection margins was also assessed by two

pT0

8 (16)

pT1

13 (26)

3 (6)

..

pT2

8 (16)

6 (11)

..

21 (42)

44 (81)

..

1 (2)

..

http://oncology.thelancet.com Vol 8 September 2007

pT2b or pT3 pT4

0

0

<0·0001

pN category, n (%) pN0

31 (62)

11 (20)

0·001

pN1

19 (38)

43 (80)

..

*Histopathological response was scored according to that described by Becker and colleagues:10 score 1a indicates complete remission and 0% residual tumour; score 1b indicates less than 10% residual tumour; score 2 indicates 10–50% residual tumour; and score 3 indicates more than 50% residual tumour.

Table 3: Surgical and histopathological outcome in patients who showed a metabolic response versus those who did not show a metabolic response to neoadjuvant chemotherapy

799

Articles

A

100 PET responders PET non-responders

Event-free survival (%)

80

60

40

20

0 0

12

24

36

Time to event (months) Number at risk 54

38

24

11

PET non-responders† 56

29

13

2

PET responders*

B

Overall survival (%)

80

60

Results

40

20

0 0

12

24

36

Survival time (months) Number at risk 54

46

30

13

PET non-responders† 56

45

21

4

Figure 2: Event-free survival and overall survival in 110 patients assessed with PET for early metabolic response to neoadjuvant chemotherapy *Four patients did not have surgery. †Two patients did not have surgery.

calculated up to the time of death from any cause or relapse, whichever occurred first.

Statistical analyses Median overall survival was the primary endpoint, assuming a ratio of metabolic responders versus nonresponders of 1:1. In previous studies, the hazard ratio (HR) for death was ≥2·0 in favour of patients who were metabolic responders after 2 weeks of chemotherapy compared with patients who were non-responders at this timepoint. The additional assumptions were: median survival of 20 months for non-responders and 40 months for responders; an accrual period of 3 years; and a maximum follow-up of 5 years. Assumptions were based on our previous data.5 In the present study, 54 patients for 800

Role of the funding source The sponsor of this study had no role in the design of the study; collection, analysis, or interpretation of the data; writing of this report; or decision to publish the results. FL, KO, BJK, TS, CP, MS, and JRS had access to the raw data. The corresponding author had full access to all of the data and the final responsibility to submit for publication.

100

PET responders*

each response stratification were necessary to detect an HR for death of 2·0 for non-responding versus responding patients with an α error of 0·05 (two-sided test) and 1-β=0·80. Secondary endpoints were the median event-free survival, postoperative complications and mortality, the number of residual tumour-free (R0) resections, and histopathological remissions. Differences in the proportions of patients were analysed using the χ² and Fisher’s exact tests. Interindividual comparisons of quantitative data were done by use of a Wilcoxon signed rank test. Survival was estimated according to Kaplan-Meier. Statistical comparisons between different groups of patients were done with a log-rank test, and HR were calculated by use of the Cox proportional hazard model. All tests were two-sided and done at the 5% level of significance with the use of SPSS for Windows, version 11.50 (SPSS Inc, Chicago, IL, USA).

Between May 27, 2002, and Aug 4, 2005, 119 consecutive patients (eight women and 111 men) were enrolled in the study. Eight of these patients were excluded because they did not meet the inclusion criteria—protocol violations were identified by one of the leading investigators (FL, KO, or JRS) within 1 week and led to exclusion from further study treatment and analysis: three patients did not have AEG type 1 or 2, but had subcardiac gastric cancer; two had haematogenous metastases; two had an inappropriate baseline SUV (SUVtumour >1·35×SUVliver+2× standard deviation SUVliver); and in one patient chemotherapy was started but, inadvertently, no baseline PET had been done. The median age of the remaining 111 patients, referred to as the per-protocol population, was 62 years (IQR 52–67). 77 patients (69%) had AEG type 1 and 34 patients (31%) had AEG type 2. Of the 111 patients, 49 (44%) were treated with cisplatin, folinic acid, and fluorouracil; 45 (41%) received additional paclitaxel, and 17 (15%) received oxaliplatin, folinic acid, and fluorouracil. Of 110 patients who were evaluable for response (one patient died before assessment), cisplatin, folinic acid, and fluorouracil were given to 26 responders and 22 non-responders; oxaliplatin, folinic acid, and fluorouracil were administered to eight responders and nine non-responders; paclitaxel, cisplatin, folinic acid, and fluorouracil were given to 20 responders and 25 nonresponders. Frequent grade 3 or 4 adverse events in 111 treated patients were as follows: diarrhoea in nine patients (all in the responder group); emesis in http://oncology.thelancet.com Vol 8 September 2007

Articles

http://oncology.thelancet.com Vol 8 September 2007

A

100

Event-free survival (%)

80

60

40

20

PET responders and histology responders PET responders and histology non-responders PET non-responders and histology non-responders

0 0

12

24

36

Time to event (months) Number at risk PET responders and histology responders

29

25

18

10

PET responders and histology non-responders

21

12

5

1

PET non-responders and histology non-responders

54

27

13

2

12

24

36

B

100

80 Overall survival (%)

seven patients (in five responders and two nonresponders); nausea in six patients (in four responders and two non-responders); and fatigue in six patients (in five responders and one non-responder). Two patients (2%) died during chemotherapy (one non-assessable patient and one responder). One of these had a sudden cardiac event, potentially induced by fluorouracil; retrospective analysis showed a skip mutation in exon 14 of the dihydropyrimidine dehydrogenase gene. The other patient had a lethal stroke of unknown relation to chemotherapy. After 2 weeks of chemotherapy, metabolic response was assessable in 110 patients (one patient died before assessment): 54 patients (49% [95% CI 39–59]) were metabolic responders and 56 patients (51% [41–61]) were metabolic non-responders. No significant differences were noted in the baseline characteristics of responders versus non-responders with regard to age, sex, performance status, tumour localisation, T-category tumour size and N-category nodal status, and histological subtype according to Laurén.12 By contrast, the tumours of metabolically responsive patients tended to be less differentiated, and these patients had significantly higher median baseline SUVs of 8·3 (IQR 6·3–11·0) versus 6·8 (IQR 5·1–9·0) in non-responders, p=0·018; table 1). The number of patients with metabolic responses were not significantly different between the different chemotherapy regimens (p=0·245; table 2). Of the 111 patients, 104 (94%) underwent surgical resection: 73 (70%; 71 with AEG type 1, two with AEG type 2) had abdominothoracic oesophagectomies, and 31 (30%, all with AEG type 2) had transhiatal extended gastrectomies. Reasons for not undergoing resection (four patients in the metabolic-responder group and two patients in the non-responder group; one patient died before assessment) were: early death (not tumour related) in two patients; tumour progression in three patients; and withdrawal of consent in two patients. The median delay between the last administration of chemotherapy and surgery was 3 weeks in the group of responding patients (range 2–5) and 2 weeks (1–4) in the group of non-responding patients. Of the 104 patients who had their tumours resected, 88 patients (85%) had tumour-free resection margins (R0 resection) and 16 patients (15%) had microscopically affected resection margins (R1). R0 resections could be done in 48 of 50 responding patients (96%) versus 40 of 54 non-responding patients (74%, p=0·002; table 3). Postoperative deaths (30-day and in-hospital mortality) occurred in two of 104 patients (2%), and postoperative complications were reported in 35 patients (34%), with no statistical difference for metabolic responders versus non-responders. In the metabolic-responder group, 29 of 50 patients (58% [95% CI 48–67]) achieved a major histopathological response (<10% residual tumour): eight patients (16%) achieved complete tumour remission; and 21 patients (42%) had subtotal remission. The SUV decrease was not

60

40

20

0 0

Survival time (months) Number at risk PET responders and histology responders

29

26

19

12

PET responders and histology non-responders

21

21

10

1

PET non-responders and histology non-responders

54

42

20

4

Figure 3: Event-free survival and overall survival according to metabolic and histopathological responses in 104 patients that underwent surgery

statistically different in patients achieving complete histological response (median decrease 56%; IQR 42–69) compared with those with subtotal histological remission (median decrease 47%; IQR 36–60). No histological response was noted in metabolic non-responders. A higher number of low-stage tumours was reported in metabolic responders than in non-responders (table 3). 801

Articles

Group, n PET responders and histology responders

29

Median event-free survival, months (95% CI)

Median overall survival, months (95% CI)

Not reached

Not reached

PET responders and histology non-responders

21

19·5 (11·8–27·2)

28·6 (22·9–34·2)

PET non-responders and histology non-responders

54

14·1 (7·5–20·6)

25·8 (19·4–32·2)

104

21·0 (14·9–27·2)

46·8 (26·2–67·3)

All patients who had surgery

Table 4: Event-free survival and overall survival according to metabolic and histopathological responses in patients who had surgery

Surviving patients were followed up for a median of 2·3 years (IQR 1·7–3·0). Per-protocol analysis of 110 patients evaluable for metabolic response showed a median event-free survival of 29·7 months (95% CI 23·6–35·7) for metabolic responders compared with 14·1 months (7·5–20·6) for non-responders (HR 2·18 [1·32–3·62], p=0·002; figure 2). Median overall survival was not reached in metabolic responders, whereas nonresponders had a median overall survival of 25·8 months (19·4–32·2; HR 2·13 [1·14–3·99], p=0·015; figure 2). Metabolic responders who also had a major histological response (n=29) had a significantly better event-free survival (HR 3·03 [1·28–7·16], p=0·006) and overall survival (HR 4·55 [1·37–15·04], p=0·004) than did metabolic responders who did not achieve a histological response (n=21). By contrast, event-free survival (HR 1·29 [0·69–2·45], p=0·430) and overall survival (HR 1·21 [0·56–2·63], p=0·549) did not differ significantly when comparing metabolic responders without a histological response (n=21) with metabolic nonresponders (n=54; figure 3; table 4). We did not note a survival difference in responding and non-responding patients who received paclitaxel-containing chemotherapy versus non-paclitaxel-containing chemotherapy (data not shown).

Discussion Our study confirms that early metabolic response measured by PET identifies patients with locally advanced adenocarcinomas of the distal oesophagus and gastric cardia (AEG type 1 and AEG type 2) who have a high chance of achieving major histological responses after neoadjuvant chemotherapy, and therefore, have a favourable prognosis. The study also shows the feasibility of a PET-response-guided treatment algorithm in the management of such tumours. Individualised, response-guided treatment concepts in oncology are needed.7 To our knowledge, our trial is the first prospective study to apply PET results from early metabolic response assessment to clinical decisionmaking in the treatment of common solid tumours. Most previously reported studies showed that changes in tumour glucose uptake that were assessed by [18F]FDGPET early, during, or immediately after neoadjuvant treatment were associated with response and prognosis.5–7,13–20 But no consequences were drawn from these findings in terms of patient management. The results of our study show that early metabolic assessment 802

can be integrated into a treatment algorithm in clinical practice. PET helps select patients who are benefiting from chemotherapy. Additionally, PET-response-guided treatment helped avoid the administration of inefficient chemotherapy to patients with no metabolic response. Locally advanced adenocarcinoma of the oesophagogastric junction is especially appropriate for this study design and objective. Our previous work has provided a sound rationale for early response assessment with PET and for devising thresholds with reproducible sensitivity and specificity for predicting traditional endpoints such as histopathological response and survival.5,6 Furthermore, the best treatment approach for this disease has not yet been defined.21–25 Therefore, a medical need exists for separating patients who eventually benefit from preoperative chemotherapy from those who might be compromised by a delay incurred through ineffective chemotherapy. The disadvantage of chemotherapy is that it can impair patients’ outcome by untoward effects and even toxic deaths, as unfortunately noted in two patients in this trial; these deaths highlight the need for confining chemotherapy to those who have a good chance of yielding a substantial benefit. The limitation of this study is its nonrandomised trial design. However, to take the relatively low incidence of adenocarcinoma of the oesophagogastric junction into consideration would have needed a multicentre setting to ensure adequate enrolment. The complexity of the technology and the absence of standardisation for metabolic imaging at the time the study was planned led us to use a single-centre, nonrandomised design. In view of the broader availability of PET machines, the maturation of the technology and validation of cut-off values, this technique could be used now in a multicentre setting. The chemotherapy regimens administered in our study show inconsistent use of paclitaxel. Previous studies defined a role for taxanes in patients with oesophagogastric cancer who are able to tolerate a three-drug regimen.26,27 When designing the protocol, we aimed to administer the most active induction chemotherapy to each patient. Our results showed no significant differences for metabolic response or survival (data not shown) in patients in the different treatment regimens. Although the subpopulations were small and different with regard to age and comorbidities, we conclude that use of PET to measure early metabolic response assists prognostication of the individual tumour biology. However, the http://oncology.thelancet.com Vol 8 September 2007

Articles

methodology of the study and the statistical power did not allow for the conclusion that PET can prognosticate independently of the chemotherapy used. One of this study’s strengths is the homogeneity of the study population. As opposed to other studies, all patients had adenocarcinomas and none had epidermoid cancer. All patients underwent sophisticated staging procedures and were assessed by a multidisciplinary treatment team before inclusion. Clinically meaningful conclusions can be drawn from this study. First, metabolic responders had a good chance of achieving complete or subtotal histological response (ie, 29 of 50 responders [58%]). This conclusion confirms our previous findings5,6 and translates into a high probability of achieving long-term survival for these patients. Second, the results suggest that early discontinuation of chemotherapy in metabolic nonresponders had no negative consequences. The primary hypothesis of this trial was that the median overall survival in metabolic responders who continue with chemotherapy will be at least twice as high compared with that of metabolic non-responders who discontinue chemotherapy after 14 days. We note that this study endpoint was statistically met, although the median follow-up time was shorter than indicated in the study protocol. But as the recorded HR for the overall survival endpoint was larger than the assumed HR underlying the protocol-specified sample-size calculation, the difference became statistically significant with even a smaller number of noted events (22% of patients had follow-ups shorter than 2 years) than was calculated in the statistical plan. In a previous study,6 we found the median recurrencefree period and median survival after continuing chemotherapy in metabolically non-responsive patients with similar tumour stages and similar preoperative chemotherapy to those in the current study were 10 months and 18 months, respectively. In the current study, the corresponding numbers for patients with no metabolic response who stopped chemotherapy were 14·1 months (95% CI 7·5–21·6) and 25·8 months (19·4–32·2). Applying all necessary caution in interpreting these results, early termination of chemotherapy based on PET results in metabolically non-responding patients did not negatively affect clinical outcome compared with previous findings. But we should emphasise that a randomised trial would be necessary to confirm this observation. PET can be used as a reference method for early response assessment in oesophagogastric cancer, and metabolic imaging has been shown to be more accurate than other technologies.28,29 While measuring changes in [18F]FDG uptake is a universal marker that correlates with the proportion of viable tumour cells, the development of new tracers specific for particular processes, such as apoptosis and proliferation, might help to achieve an even more accurate prediction of response, especially http://oncology.thelancet.com Vol 8 September 2007

when integrating biologically targeted agents into neoadjuvant treatment.30 Our trial showed that patients who respond to neoadjuvant chemotherapy who are identified by early metabolic imaging have a favourable prognosis, and this is especially true for metabolic responders who also achieve a major histopathological response. However, patients who do not achieve a histological response, despite previous metabolic response (in our study 20% of those patients who underwent resection), the prognosis remains dismal. Therefore, histological response remains an important prognosticator that seems to be stronger and more robust than early metabolic response. The cellular mechanisms for the observation that a subset of tumours have early response by PET, but not by histology, is currently unknown. Importantly, histological response can only be determined after neoadjuvant treatment and resection. This timepoint would be appropriate for the planning of adjuvant treatment strategies, but it definitely comes too late for tailoring of neoadjuvant treatment. The important effect of metabolic imaging in this context is that PET can predict histological response earlier and with higher accuracy than any other clinical assessment.28,29 Although histological responses might be detectable by sequential endoscopic biopsies, histological examination of resection specimens has shown that in neoadjuvanttreated oesophagogastric cancer, residual tumour is often located in the outward layers of the intestinal wall, while the luminal parts have become fibrotic and are free of residual tumour.10 Therefore, the endoscopic approach cannot rule out residual viable tumour. A higher metabolic cut-off value might be even better able to predict histopathological response. Indeed, a previous study5 from our group showed that a 45% or more decrease in SUV would result in higher specificity for histological response (86% vs 75%) but in a lower negative predictive value than the cut-off level used in the current study (35% or more decrease in SUV). In clinical terms, this translates into a bigger proportion of patients from whom chemotherapy would be withheld, despite having chemosensitive tumours. For that reason, our group adhered to the 35% cut-off value in all consecutive studies.6,31 The cut-off value of 35% was also clearly stated in the current protocol. The consistency of the positive predictive values of the PET measurements for predicting metabolic response noted in this trial (58% [95% CI 47–69]) compared with our previous trials (53% in the first study,5 44% in the second study6) justifies this decision. In other clinical settings—eg, chemoradiation in oesophageal squamous-cell cancer and rectal cancer— different cut-offs have proved to be more accurate.32,33 In the current study, 23 of the 54 metabolic responders had a decrease in SUV in the range of 35–45% 2 weeks after the start of chemotherapy. Of these 23 patients, ten had a major histopathological response after full chemotherapy. If the cut-off value had been 45%, these patients would not have received efficient neoadjuvant treatment. 803

Articles

The feasibility of PET-guided neoadjuvant treatment and thresholds for a decrease in the SUV have now been established. Additionally, a hypothesis has been generated by this trial that early termination of chemotherapy based on PET does not negatively affect clinical outcome in metabolic non-responders. But important issues remain to be addressed in future trials. A future phase III trial should randomise metabolic non-responders between immediate surgery and 12 weeks of chemotherapy followed by surgery. Also, the best duration of chemotherapy needs to be established. In this study, we routinely administered a 12-week chemotherapy regimen to patients responding metabolically before they proceeded to resection. A shorter regimen of 6–9 weeks might also be sufficient for achieving maximum tumour shrinkage, thereby helping to achieve an R0 resection.1,2 Furthermore, the role of surgery in patients responding to preoperative treatment needs clarification. If complete histopathological responses can be predicted preoperatively, these patients might be candidates for organ preservation. However, currently available methods of clinical investigation do not enable precise identification of complete histological remission. The present and previous studies failed to show a significant difference in the amount of SUV decrease shown by patients who achieved complete remission and those whose histological remission was subtotal.6 Furthermore, the results of salvage oesophagectomy in patients with subsequent tumour progression have not been convincing thus far.34 Nonetheless, first observation studies suggest a potential role for salvage surgery, at least in oesophageal squamous cancer, if the tumour height and aortic contact are limited.35 However, for locally advanced adenocarcinomas, surgery after neoadjuvant treatment remains the mainstay of treatment.36 The value of the pretherapeutic SUV is also under debate. While most studies did not attribute a significant prognostic value to pretherapeutic SUV,6,15–17 others did.18,37 In our study, metabolic responders and non-reponders showed a slightly different baseline SUV. But a clear cut-off value separating responding from non-responding patients was not detectable. In view of potential correlations between baseline SUVs and prognosis or response to neoadjuvant treatment, selection criteria strictly prohibited the enrolment of patients with very low tumour SUVs. Another important question is whether tumours that do not respond to conventional chemotherapy might eventually respond if radiation or biological responsemodifiers, such as monoclonal antibodies or tyrosine kinase inhibitors, are added. Preoperative chemoradiation has been studied in locally advanced adenocarcinoma of the oesophagogastric junction and it might be another valuable option.38–40 Because greater histopathological responses have been recorded after neoadjuvant chemoradiation compared with after chemotherapy alone,3 a stronger benefit with chemoradiation might be suggested. However, results from currently available phase III trials suggest that neoadjuvant chemotherapy alone is the best 804

treatment option.1,2,25 Of note, if resection is done after neoadjuvant chemoradiation, increased surgical morbidity and mortality might be expected.41,42 Further investigations are needed to find out whether patients who do not respond to chemotherapy might be suitable candidates for radiation treatment in addition to chemotherapy. The antitumoral activity of chemoradiation could also be monitored by early PET assessments,17,20 thus encouraging new innovative study designs. With the advent of new cytotoxic and biologically targeted agents, such as monoclonal antibodies or tyrosine kinase inhibitors, with proven—or at least assumed—activity against oesophagogastric adenocarcinomas, other treatment regimens could also be studied in patients with primarily chemoresistant tumours.43 However, early metabolic response evaluation might unmask tumours with an unfavourable biology and an irreversibly poor prognosis, regardless of the treatment applied. In summary, this trial delineates how response-guided treatment algorithms might be applied to clinical practice. Doing so could provide a model for other malignant diseases, such as lung, head and neck, or ovarian cancer, for which induction treatment has a potential role. This hypothesis needs to be addressed in randomised phase III trials. Contributors FL, KO, WAW, HJS, UF, CP, MS, and JRS were responsible for the study conception and design. FL, B-JK, RB, HH, and JRS provided administrative support. FL, SL, HW, KH, and RB collected and assembled the data. FL, KO, B-JK, HJS, TS, HH, UF, CP, MS, and JR analysed and interpreted the data. FL, B-JK, WAW, TS, CP, MS, and JRS wrote the report. All authors approved the final report. Conflicts of interest The authors declared no conflicts of interest. Acknowledgments This study was supported by a grant from the Commission for Clinical Research of the Technische Universität München (grant number KKF 01-03). The authors gratefully acknowledge the assistance of the PET technologists, chemists, and staff of the Munich Center for Clinical Studies. References 1 Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 2006; 80: 11–20. 2 Medical Research Council Oesophageal Cancer Working Party. Surgical resection with or without preoperative chemotherapy in oesophageal cancer: a randomised controlled trial. Lancet 2002; 359: 1727–33. 3 Gebski V, Burmeister B, Smithers BM, Foo K, Zalcberg J, Simes J. Survival benefits from neoadjuvant chemoradiotherapy or chemotherapy in oesophageal carcinoma: a meta-analysis. Lancet Oncol 2007; 8: 226–34. 4 Kelsen DP, Ginsberg R, Pajak TF, et al. Chemotherapy followed by surgery compared with surgery alone for localized esophageal cancer. N Engl J Med 1998; 339: 1979–84. 5 Weber WA, Ott K, Becker K, et al. Prediction of response to preoperative chemotherapy in adenocarcinomas of the esophagogastric junction by metabolic imaging. J Clin Oncol 2001; 19: 3058–65. 6 Ott K, Weber WA, Lordick F, et al. Metabolic imaging predicts response, survival, and recurrence in adenocarcinomas of the esophagogastric junction. J Clin Oncol 2006; 24: 4692–98. 7 Juweid ME, Cheson BD. Positron-emission-tomography and assessment of cancer therapy. N Engl J Med 2006; 354: 496–507.

http://oncology.thelancet.com Vol 8 September 2007

Articles

8 9

10

11 12

13

14

15

16

17

18

19

20

21 22 23

24

25 26

Siewert JR, Stein HJ. Classification of adenocarcinoma of the oesophagogastric junction. Br J Surg 1998; 85: 1457–59. Bartels H, Stein HJ, Siewert JR. Preoperative risk analysis and postoperative mortality of oesophagectomy for resectable oesophageal cancer. Br J Surg 1998; 85: 840–44. Becker K, Mueller JD, Schulmacher C, et al. Histomorphology and grading of regression in gastric carcinoma treated with neoadjuvant chemotherapy. Cancer 2003; 98: 1521–30. Sarbia M, Becker KF, Hofler H. Pathology of upper gastrointestinal malignancies. Semin Oncol 2004; 31: 465–75. Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal type carcinoma: an attempt at a histo-clinical classification. Acta Pathol Microbiol Scand 1965; 64: 31–43. Brucher BL, Weber W, Bauer M, et al. Neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission tomography. Ann Surg 2001; 233: 300–09. Flamen P, Van Cutsem E, Lerut A, et al. Positron emission tomography for assessment of the response to induction radiochemotherapy in locally advanced oesophageal cancer. Ann Oncol 2002; 13: 361–68. Downey RJ, Akhurst T, Ilson D, et al. Whole body 18FDG-PET and the response of esophageal cancer to induction therapy: results of a prospective trial. J Clin Oncol 2003; 21: 428–32. Swisher SG, Maish M, Erasmus JJ, et al. Utility of PET, CT, and EUS to identify pathologic responders in esophageal cancer. Ann Thorac Surg 2004; 78: 1152–60. Wieder HA, Brucher BL, Zimmermann F, et al. Time course of tumor metabolic activity during chemoradiotherapy of esophageal squamous cell carcinoma and response to treatment. J Clin Oncol 2004; 22: 900–08. Levine EA, Farmer MR, Clark P, et al. Predictive value of 18-fluorodeoxy-glucose-positron emission tomography (18F-FDG-PET) in the identification of responders to chemoradiation therapy for the treatment of locally advanced esophageal cancer. Ann Surg 2006; 243: 472–78. Gillham CM, Lucey JA, Keogan M, et al. (18)FDG uptake during induction chemoradiation for oesophageal cancer fails to predict histomorphological tumour response. Br J Cancer 2006; 95: 1174–79. Westerterp M, Omloo JM, Sloof GW, et al. Monitoring of response to pre-operative chemoradiation in combination with hyperthermia in oesophageal cancer by FDG-PET. Int J Hyperthermia 2006; 22: 149–160. Enzinger PC, Mayer RJ. Esophageal cancer. N Engl J Med 2003; 349: 2241–52. Ilson DH. Cancer of the gastroesophageal junction: combined modality therapy. Surg Oncol Clin N Am 2006; 15: 803–24. Das P, Fukami N, Ajani JA. Combined modality therapy of localized gastric and esophageal cancers. J Natl Compr Canc Netw 2004; 4: 375–82. Chong G, Cunningham D. The role of preoperative chemotherapy for esophageal cancer: when is surgery alone feasible? Nat Clin Pract Oncol 2005; 2: 172–73. Lordick F, Stein HJ, Peschel C, Siewert JR. Neoadjuvant therapy for oesophagogastric cancer. Br J Surg 2004; 91: 540–51. Ilson DH, Ajani J, Bhalla K, et al. Phase II trial of paclitaxel, fluorouracil, and cisplatin in patients with advanced carcinoma of the esophagus. J Clin Oncol 1998; 16: 1826–34.

http://oncology.thelancet.com Vol 8 September 2007

27

28

29

30 31

32

33

34

35

36

37

38

39

40

41

42

43

Van Cutsem E, Moiseyenko VM, Tjulandin S, et al. Phase III study of docetaxel and cisplatin plus fluorouracil compared with cisplatin and fluorouracil as first-line therapy for advanced gastric cancer: a report of the V325 Study Group. J Clin Oncol 2006; 24: 4991–97. Westerterp M, van Westreenen HL, Reitsma JB, et al. Esophageal cancer: CT, endoscopic US, and FDG PET for assessment of response to neoadjuvant therapy—systematic review. Radiology 2005; 236: 841–51. Wieder HA, Beer AJ, Lordick F, et al. Comparison of changes in tumor metabolic activity and tumor size during chemotherapy of adenocarcinomas of the esophagogastric junction. J Nucl Med 2005; 46: 2029–34. Weber WA. Positron emission tomography as an imaging biomarker. J Clin Oncol 2006; 24: 3282–92. Ott K, Fink U, Becker K, et al. Prediction of response to preoperative chemotherapy in gastric carcinoma by metabolic imaging: results of a prospective trial. J Clin Oncol 2003; 21: 4604–10. Brucher BL, Weber W, Bauer M, et al. Neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission tomography. Ann Surg 2001; 233: 300–09. Cascini GL, Avallone A, Delrio P, et al. 18F-FDG PET is an early predictor of pathologic tumor response to preoperative radiochemotherapy in locally advanced rectal cancer. J Nucl Med 2006; 47: 1241–48. Swisher SG, Wynn P, Putnam JB, et al. Salvage esophagectomy for recurrent tumors after definitive chemotherapy and radiotherapy. J Thorac Cardiovasc Surg 2002; 123: 175–183. Piessen G, Briez N, Triboulet JP, Mariette C. Patients with locally advanced esophageal carcinoma nonresponder to radiochemotherapy: who will benefit from surgery? Ann Surg Oncol 2007; 14: 2036–44. Mariette C, Piessen G, Triboulet JP. Therapeutic strategies in oesophageal carcinoma: role of surgery and other modalities. Lancet Oncol 2007; 8: 545–53. Choi JY, Jang HJ, Shim YM, et al. 18F-FDG PET in patients with esophageal squamous cell carcinoma undergoing curative surgery: prognostic implications. J Nucl Med 2004; 45: 1843–50. Walsh TN, Noonan N, Hollywood D, Kelly A, Keeling N, Hennessy TP. A comparison of multimodal therapy and surgery for esophageal adenocarcinoma. N Engl J Med 1996; 335: 462–67. Urba SG, Orringer MB, Turrisi A, et al. Randomized trial of preoperative chemoradiation versus surgery alone in patients with locoregional esophageal carcinoma. J Clin Oncol 2001; 19: 305–13. Tepper JE, Krasna M, Niedzwicki D, et al. Superiority of trimodality therapy to surgery alone in esophageal cancer: results of CALGB 9781. J Clin Oncol 2006; 24 (suppl 18S): 4012. Fiorica F, Di Bona D, Schepis F, Iannettoni M, Forestiere A, Strawderman M. Preoperative chemoradiotherapy for oesophageal cancer: a systematic review and meta-analysis. Gut 2005; 53: 925–30. Steyerberg EW, Neville BA, Koppert LB, et al. Surgical mortality in patients with esophageal cancer: development and validation of a simple risk score. J Clin Oncol 2006; 24: 4277–84. Ilson D, Bains M, Rizk N, et al. Phase II trial of preoperative cisplatin-irinotecan followed by concurrent cisplatin-irinotecan and radiotherapy: PET scan after induction therapy may identify early treatment failure. J Clin Oncol 2006; 24 (suppl 18S): 4023.

805