IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer Adjuvant chemotherapy in rectal cancer: Defining subgroups who may benefit after neoadjuvant chemoradiation and resection A pooled analysis of 3,313 patients Monique Maas 1,2, Patty J Nelemans 3, Vincenzo Valentini 4, Christopher H. Crane 5, Carlo Capirci 6, Claus R odel 7, Garrett M. Nash 8, Li-Jen Kuo 9, Rob Glynne-Jones 10, Julio Garcıa-Aguilar 8, Javier Suarez 11, Felipe A. Calvo 12, Salvatore Pucciarelli 13, Sebastiano Biondo 14, George Theodoropoulos 15, Doenja M.J. Lambregts 2, Regina G.H. Beets-Tan 2 and Geerard L. Beets 1 1 Department of Surgery, Maastricht University Medical Centre, Maastricht, The Netherlands 2 Department of Radiology, Maastricht University Medical Centre, Maastricht, The Netherlands 3 Department of Epidemiology, Maastricht University, Maastricht, The Netherlands 4 Department of Radiation Oncology, Universita Cattolica S. Cuore, Policlinico Universitario Agostino Gemelli, Largo A. Gemelli, 8 Rome 00168, Italy 5 Department of Radiation Oncology, University of Texas MD Anderson Cancer Centre, Houston, TX Department of Radiotherapy, State Hospital, Rovigo, Italy 7 Department of Radiation Oncology, Universit atsklinikum Frankfurt, Frankfurt am Main, Germany 8 Department Of Surgery, Memorial Sloan-Kettering Cancer Centre, New York, NY Department of General and Colorectal Surgery, Taipei Medical University Hospital, Graduate Institute of Clinical Medicine, College of Medicine, Taipei Medical University, Taipei 11031, Taiwan 10 Department of Clinical Oncology, Mount Vernon Hospital, HA6 2RN London, United Kingdom 11 Department of Surgery, Colorectal Unit Complejo Hospitalario de Navarra, Pamplona, Spain 12 Department of Oncology, General University Hospital Gregorio Mara~non, Madrid 28007, Spain 13 Department of Oncological and Surgical Sciences, Surgical Clinic 2nd, University of Padova, Padova, Italy 14 Department of Surgery, Bellvitge University Hospital, IDIBELL, University of Barcelona, Barcelona, Spain 15 First department of Propaedeutic Surgery, Athens Medical School, Hippocration General Hospital, GR Athens, Greece Recent literature suggests that the benefit of adjuvant chemotherapy (act) for rectal cancer patients might depend on the response to neoadjuvant chemoradiation (CRT). Aim was to evaluate whether the effect of act in rectal cancer is modified by response to CRT and to identify which patients benefit from act after CRT, by means of a pooled analysis of individual patient data from 13 datasets. Patients were categorized into three groups: pcr (ypt0n0), ypt1-2 tumour and ypt3-4 tumour. Hazard ratios (HR) for the effect of act were derived from multivariable Cox regression analyses. Primary outcome measure was recurrence-free survival (RFS). One thousand seven hundred and twenty three (1723) (52%) of 3,313 included patients received act. Eight hundred and ninety eight (898) patients had a pcr, 966 had a ypt1-2 tumour and 1,302 had a ypt3-4 tumour. For 122 patients response, category was missing and 25 patients had ypt0n1. Median follow-up for all patients was 51 (0 219) months. HR for RFS with 95% CI for patients treated with act were 1.25( ), 0.58( ) and 0.83( ) for patients with pcr, ypt1-2 and ypt3-4 tumours, respectively. The effect of act in rectal cancer patients treated with CRT differs between subgroups. Patients with a pcr after CRT may not benefit from act, whereas patients with residual tumour had superior outcomes when act was administered. The test for interaction did not reach statistical significance, but the results support further investigation of a more individualized approach to administer act after CRT and surgery based on pathologic staging. Key words: rectal cancer, neoadjuvant treatment, adjuvant chemotherapy, survival, response Abbreviatons: act: adjuvant chemotherapy; CRT: chemoradiation; 95% CI: 95% confidence interval; DFS: disease free survival; HR: hazard ratio; OS: overall survival; pcr: pathologic complete response; RCT: randomized controlled trial; RFS: recurrence free survival There are no acknowledgements for this paper. DOI: /ijc History: Received 13 Mar 2014; Accepted 7 Oct 2014; Online 22 Nov 2014 Correspondence to: G.L. Beets, Dept of Surgery, Maastricht University Medical Centre, PO Box 5800, 6202 AZ Maastricht, The Netherlands, Tel.: , g.beets@mumc.nl Int. J. Cancer: 137, (2015) VC 2014 UICC

2 Maas et al. 213 What s new? Most patients with locally advanced rectal cancer receive adjuvant chemotherapy after neoadjuvant treatment and surgery. Based on a pooled analysis of individual patient data from 13 datasets, this study however shows that the benefit of adjuvant chemotherapy differs between subgroups, based on the response of patients to previous neoadjuvant chemoradiation. Patients with a complete response after chemoradiation may not benefit from adjuvant chemotherapy, whereas patients with residual tumour have superior outcomes when adjuvant chemotherapy was administered. The results support further investigation of a more individualized approach based on pathologic staging for the administration of adjuvant chemotherapy after chemoradiation and surgery. Currently, adjuvant chemotherapy (act) is indicated for colorectal cancer patients at increased risk for metastatic disease, typically established through postoperative histopathological risk factors like involved lymph nodes, poor differentiation grade or lymphatic or venous invasion. 1 For rectal cancer, the shift from adjuvant to neoadjuvant (chemo)radiation 2 has complicated the use of postoperative histopathology to determine the indication for act, as chemoradiation (CRT) often leads to downstaging of tumour and nodes and the histopathological staging may no longer reflect the initial stage. The pre-crt baseline imaging could be used for decision making, but the available imaging modalities are only moderately accurate in lymph node assessment, 3,4 a risk factor, that is, considered pivotal in decision making about act. Another option is to guide the decision on post-crt histopathological staging, but this strategy carries the risk of undertreating patients whose lymph nodes have been sterilized. A highly pragmatic approach is to consider all tumors that receive neoadjuvant CRT also at risk for distant recurrence and treat accordingly with act. However, this strategy carries the risk of overtreatment, as some distally located smaller tumors are also treated with neoadjuvant chemoradiation. What complicates this issue further is the fact that the evidence for survival benefit of act is weaker in rectal cancer than for colonic cancer, with conflicting results from meta-analyses and some experts questioning its routine role in modern rectal cancer management. 5,6 The question remains as to which subgroups of patients could benefit from act after CRT for rectal cancer. 7 An early subgroup analysis of a randomized EORTC trial suggested a beneficial effect of act for patients with ypt0-2 tumours, not seen in ypt3-4 tumours. 8 The significant difference between both response groups was no longer observed after long-term follow-up. Another report questioned the benefit of act for pathologic complete responders (ypt0n0) after CRT. 9 Therefore, the aim of this study was to evaluate whether the effect of act after CRT is modified by the histopathological stage after CRT using a pooled analysis of individual patient data. Material and Methods Individual patient data from a previous pooled analysis on the prognostic value of a pcr after CRT for locally advanced rectal cancer were used. For that pooled analysis, studies were included when they evaluated patients with primary locally advanced rectal cancer that were treated with CRT and reported long-term outcome and response as well as patient and treatment characteristics. More details on the search strategy can be found in an earlier publication. 10 From the primary literature search, 17 articles were identified of which 14 authors consented to participation in the previous analysis. For the present analysis, information was needed on the administration of act and/or about the ypt and ypnstage. Such information was missing in three of these 14 datasets and these three datasets were not included for the current analysis From one of the three datasets that could not be included at the time of the earlier publication, data was available for the current study. 14,15 Additionally, to increase the sample of patients with a pcr after CRT,- we contacted the first author of a study on the long-term outcome of a large group of patients with a pcr after CRT, who agreed to share the dataset. 9 Thus, data from 13 centers were included for the present analysis. 9,14 26 Five studies were prospective 17,23,27 29 and eight were retrospective, 9,18,20 22,24,25,30,31 for one dataset data was derived from an RCT. 23 Baseline staging of patients before treatment consisted of endorectal ultrasound and computed tomography mostly. Interval between CRT and surgery was generally 6 8 weeks. Surgery consisted of total mesorectal excision for all patients. Individual patients were excluded from the current pooled analysis for the following reasons: (i) data about administration of act was missing, (ii) patients were treated with preoperative radiation only without concomitant chemotherapy, (iii) radiotherapy dose 25 Gy, (iv) presence of synchronous metastasis, (v) locally recurrent disease or (vi) an incomplete resection (R1 or R2). Included for analysis were data on gender, age, clinical T- and N-stage, distance between tumour and anal verge, pathologic T- and N- stage, type of surgery (low anterior resection, abdominoperineal resection or other surgery, for example, Hartmann s procedure or exenteration), type of act and information on dates of events and/or follow-up times to events. Based on histopathology of the resection specimen patients were categorized into: (i) pcr, defined as ypt0n0, (ii) ypt1-2 and (iii) ypt3-4. For an additional analysis to evaluate the effect of act in patients with ypn0 and ypn1, patients were categorized into subgroups according to pathological N-status (ypn-stage). For 147 patients (4%), the

3 Table 1. Overview of baseline characteristics per included study Valentini Crane Capirci R odel Nash Kuo Glynne-Jones Garcıa-Aguilar Suarez Calvo Pucciarelli Biondo Theodoropoulos N Study type Prospective Prospective (RCT arm) Prospective Prospective Prospective Age Gender (male) 61% (357/582) 63% (326/517) 67% (293/439) 70% (246/350) 66% (216/329) 59% (145/245) 70% (112/159) 55% (86/156) 70% (83/119) 67% (78/117) 63% (67/106) 73% (75/103) 72% (63/88) Missing 0% 0% 1% (2/441) 0% 0% 0% 1% (1/160) 0% 0% 0% 0% 0% 0% Follow-up 50 (1 219) 56 (0 168) 39 (0 154) 92 (3 156) 69 (5 188) 33 (1 121) 37 (0 140) 32 (1 111) 42 (3 86) 71 (0 143) 61 (9 141) 41 (0 94) 33 (2 95) ct-stage ct1 1% (4/512) 0% (1/509) 0% (1/438) 0% 0% (1/295) 0% 0% 1% (2/156) 0% 0% 0% 0% 0% ct2 4% (19/512) 6% (30/509) 6% (28/438) 5% (15/280) 11% (31/295) 7% (16/245) 4% (6/158) 5% (8/156) 7% (8/118) 3% (3/116) 11% (11/104) 2% (2/103) 0% ct3 82% (420/512) 85% (433/509) 85% (370/438) 87% (244/280) 84% (249/295) 85% (207/245) 49% (77/158) 83% (130/156) 88% (104/118) 87% (101/116) 70% (73/104) 92% (95/103) 94% (83/55) ct4 14% (69/512) 9% (45/509) 9% (39/438) 8% (21/280) 5% (14/295) 9% (22/145) 47% (75/158) 10% (16/156) 5% (6/118) 10% (12/116) 19% (20/104) 6% (6/103) 6% (5/88) Missing 12% (70/582) 2% (8/517) 1% (3/441) 20% (70/350) 10% (34/329) 0% 1% (2/160) 0% 1% (1/119) 1% (1/117) 2% (2/106) 0% 0% cn-stage cn1 76% (419/551) 54% (277/509) 40% (169/418) 18% (59/329) 54% (137/252) 86% (182/212) 53% (82/156) 72% (112/156) 42% (50/118) 48% (54/113) 65% (67/103) 81% (79/98) 48% (42/88) Missing 5% (31/582) 2% (8/517) 5% (23/441) 6% (21/350) 23% (77/329) 14% (33/245) 3% (4/160) 0% 1% (1/119) 3% (4/117) 3% (3/106) 5% (5/103) 0% Distance from anal verge NR NR 1 5 cm 65% (317/485) 52% (267/517) 44% (150/338) 40% (131/329) 50% (123/245) 57% (91/159) 37% (57/156) 48% (57/119) 43% (45/105) 42% (45/106) 55% (50/91) > 5 cm 35% (168/485) 48% (250/517) 56% (188/338) 60% (198/329) 50% (122/245) 43% (68/159) 64% (99/156) 52% (62/119) 57% (60/105) 58% (61/106) 45% (41/91) Missing 17% (97/582) 0% 3% (12/350) 0% 0% 1% (1/160) 0% 0% 10% (12/117) 0% 12% (12/103) ypt-stage ypt0 20% (98/486) 22% (111/512) 100% (441/441) 11% (38/348) 22% (72/329) 17% (42/242) 22% (34/158) 15% (24/156) 14% (17/119) 9% (10/117) 18% (19/106) 16% (16/103) 19% (17/88) ypt1 6% (31/486) 6% (32/512) 0% 6% (22/348) 8% (26/329) 3% (7/242) 4% (6/158) 5% (8/156) 4% (5/119) 9% (11/117) 8% (8/106) 6% (6/103) 1% 1(1/88) ypt2 28% (138/486) 30% (152/512) 0% 31% (109/348) 32% (106/329) 26% (63/242) 15% (24/158) 30% (46/156) 25% (30.119) 34% (40/117) 43% (45/106) 31% (32/103) 21% (18/88) ypt3 42% (205/486) 37% (189/512) 0% 49% (169/348) 38% (124/329) 51% (124/242) 53% (84/158) 48% (75/156) 53% (63/119) 39% (46/117) 31% (33/106) 47% (48/103) 55% (48/88) ypt4 3% (16/486) 6% (28/512) 0% 3% (10/348) 0% (1/329) 2% (6/242) 6% (10/158) 2% (3/156) 3% (4/119) 9% (10/117) 1% (1/106) 1% (1/103) 5% (4/88) Missing 17% (96/582) 1% (5/517) 0% 1% (2/350) 0% 1% (3/245) 1% (2/158) 0% 0% 0% 0% 0% 0% ypn-stage ypn1 25% (138/555) 25% (123/492) 0% 30% (104/350) 0% 33% (80/245) 25% (37/147) 31% (47/154) 24% (28/119) 20% (23/117) 19% (20/106) 25% (26/103) 33% (29/88) Missing 5% (27/582) 5% (25/517) 0% 0% 0% 0% 8% (13/160) 1% (2/156) 0% 0% 0% 0% 0% Type of surgery LAR 69% (400/582) 66% (343/517) 77% (339/441) 68% (238/350) 86% (282/329) 82% (202/245) 38% (60/160) 60% (57/156) 74% (88/119) 59% (69/117) 87% (92/106) 60% (62/103) 64% (56/88)

4 Table 1. Overview of baseline characteristics per included study (Continued) Valentini Crane Capirci R odel Nash Kuo Glynne-Jones Garcıa-Aguilar Suarez Calvo Pucciarelli Biondo Theodoropoulos APR 18% (103/582) 28% (145/517) 23% (100/441) 25% (89/350) 14% (47/329) 14% (34/245) 59% (95/160) 38% (59/156) 24% (28/119) 41% (48/117) 12% (13/106) 40% (41/103) 36% (32/88) Other 14% (79/582) 6% (29/517) 1% (2/441) 7% (23/350) 0% 4% (9/245) 3% (5/160) 3% (4/156) 3% (3/119) 0% 1% (1/106) 0% 0% Missing 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Response category ypt0n0 (pcr) 20% (95/485) 18% (90/491) 100% (441/441) 10% (36/346) 22% (72/329) 15% (36/236) 18% (28/152) 14% (21/153) 14% (17/119) 9% (10/117) 18% (19/106) 16% (16/103) 19% (17/88) ypt1-2 35% (169/485) 38% (184/491) 0% 38% (131/346) 40% (132/329) 30% (70/236) 20% (30/152) 35% (54/153) 29% (35/119) 44% (51/117) 50% (53/106) 37% (38/103) 22% (19/88) ypt3-4 46% (221/485) 44% (217/491) 0% 52% (179/346) 38% (125/329) 55% (130/236) 62% (94/152) 51% (78/153) 56% (67/119) 48% (56/117) 32% (34/106) 48% (49/103) 59% (52/88) Missing 17% (97/582) 5% (26/517) 0% 1% (4/350) 0% 4% (9/245) 5% (8/160) 2% (3/156) 0% 0% 0% 0% 0% Adjuvant chemotherapy Yes 18% (107/582) 73% (375/517) 18% (81/441) 85% (296/350) 83% (272/329) 60% (147/245) 19% (31/160) 21% (32/156) 82% (97/119) 62% (73/117) 57% (60/106) 100% (103/103) 56% (49/88) Type of chemotherapy 5-FU 5-FU and LV 5-FU and LV or FA or FOLFOX 5-FU 5-FU and LV/OX/IRIN 5-FU and LV 1/2 levamisole) 5-FU and LV 5-FU and LV 5-FU or CAP or XELOX 5-FU and LV 5-FU and LV 5-FU and LV 5-FU and LV 1 For patients from this study, distance from the anal verge was categorized in 7cm or >7cm and therefore distance from anal verge could not be retrieved. LAR5low anterior resection; APR5abdominoperineal resection; pcr5pathologic complete response (5ypT0N0). NR5not reported. 5-FU55-fluorouracil; LV5leucovorin; FA5folic acid; FOLFOX5folic acid with 5-fluorouracil and oxaliplatin; OX5oxaliplatin; IRIN5irinotecan; CAP5capecitabine; XELOX5capecitabine with oxaliplatin.

5 216 Adjuvant chemotherapy in rectal cancer Table 2. Baseline characteristics for all patients and for patients categorized according to administration of act All (n patients) response category was missing (n 5 122) or was not consistent with one of the abovementioned categories (n 5 25, ypt0n1). These data, however, were included in the analysis to evaluate the effect of act in the total group of patients. The primary outcome measure was recurrence-free survival (RFS). Recurrence was defined as a local and/or distant recurrence during follow-up. No adjuvant chemotherapy (n 5 1,590) Adjuvant chemotherapy (n 5 1,723) p-value Age (years) 61 (12) 63 (11) 60 (12) p < Gender (male) 64% (2,119/3,313) 63% (1,001/1,590) 65% (1,118/1,723) Missing 0% (3/3,298) 0% (2/1,575) 0% (1/1,723) Clinical T-stage ct1 0% (9/3,122) 1% (7/1,489) 0% (2/1,618) ct2 6% (177/3,122) 6% (93/1,489) 5% (84/1,618) ct3 83% (2,586/3,122) 81% (1,225/1,489) 84% (1,361/1,618) ct4 11% (350/3,122) 12% (179/1,504) 11% (171/1,618) Missing 6% (191/3,313) 5% (86/1,590) 6% (105/1,723) Clinical N-stage cn1 56% (1,729/3,103) 56% (846/1,508) 55% (883/1,595) Missing 6% (210/3,313) 5% (82/1,590) 7% (128/1,723) Distance from anal verge (cm) 5 cm 50% (1,333/2,650) 55% (606/1,099) 47% (727/1,551) < > 5 cm 50% (1,317/2,650) 45% (493/1,099) 53% (824/1,551) Missing 20% (663/3,313) 31% (491/1,590) 10% (172/1,723) Type of surgery LAR 70% (2,324/3,313) 68% (1,073/1,590) 73% (1,251/1,723) APR 25% (834/3,298) 26% (416/1,590) 24% (418/1,723) < Other 5% (155/3,298) 6% (101/1,590) 3% (54/1,723) Missing 0% 0% 0% Response category pcr 28% (898/3,166) 41% (608/1,478) 17% (290/1,688) ypt1-2 31% (966/3,166) 27% (400/1,478) 34% (566/1,688) < ypt3-4 41% (1,302/3,166) 32% (470/1,478) 49% (832/1,688) Missing 4% (147/3,313) 7% (112/1,590) 2% (35/1,723) Pathological N-stage N1 20% (655/3,246) 13% (194/1,538) 27% (461/1,708) < Missing 2% (67/3,313) 3% (52/1,590) 1% (15/1,723) Median follow-up time (months) 51 (0 219) 47 (0 219) 55 (0 188) < cm 5 centimetre; LAR 5 low anterior resection; APR 5 abdominoperineal resection; pcr 5 pathologic complete response (5ypT0N0). Statistical analysis Baseline characteristics were compared between patients who did and did not receive act. Differences in baseline characteristics were tested for statistical significance with the independent samples t-test for the comparison of means and the v 2 -test for comparison of proportions. Outcome measures were RFS, defined as the absence of any recurrence (local or distant) during follow-up; disease-free survival (DFS), defined as the absence of both local and distant recurrence and death by any cause during follow-up; and overall survival (OS), defined as the absence of death from any cause during follow-up. Separate analyses were performed for local control (defined as the absence local recurrence and death by any cause during followup) or distant control (defined as the absence distant recurrence and death by any cause during follow-up). Follow-up time was calculated from the date of surgery to the date of the event of interest or the date of last follow-up (censoring date). Multivariable Cox proportional hazards models with stratification for study center were used to adjust for differences in baseline characteristics between patients who did and did not receive act within response groups. The proportional hazard assumption was evaluated graphically with log-minus-log

6 Maas et al. 217 Table 3. Baseline characteristics for patients categorized according to their response to chemoradiation pcr (n 5 898) ypt1-2 (n 5 966) ypt3-4 (n 5 1,302) p-value Age (years) 61 (611) 61 (611,5) 61 (612,0) p Gender (male) 66% (589/896) 65% (632/966) 65% (843/1,302) p Missing 0.2% (2/898) 0% 0% (1/1,302) Clinical T-stage ct1 0.5% (4/878) 0.3% (3/924) 0.1% (1/1,215) ct2 8% (71/878) 9% (81/924) 2% (21/1,215) ct3 82% (723/878) 84% (776/924) 82% (999/1,215) p < ct4 9% (80/878) 6% (64/924) 16% (194/1,215) Missing 2% (20/898) 4% (42/966) 7% (87/1,302) Clinical N-stage cn1 50% (431/855) 55% (501/907) 59% (486/1,198) p < Missing 5% (43/898) 6% (59/966) 8% (104/1,302) Tumour height (cm) 5 cm 53% (230/434) 54% (491/912) 45% (545/1,201) >5 cm 47% (204/434) 46% (421/912) 55% (656/1,201) p < Missing 52% (464/898) 6% (54/966) 8% (102/1,302) Type of surgery LAR 76% (680/898) 71% (686/966) 68% (882/1,302) APR 22% (196/898) 25% (238/966) 28% (367/1,302) p Other 2% (22/898) 4% (42/966) 4% (53/1,302) Missing 0% 0% 0% pcr 5 pathologic complete response; RFS 5 relapse-free survival; DFS 5 disease-free survival; OS 5 overall survival. Table 4. Five-year Kaplan Meier estimates (with 95% CI) of patients with varying response to CRT with regard to T- and N-stage who were not treated with act Recurrence free survival Disease-free survival Overall survival pcr 90% (87 93) 87% (84 90) 87% (83 90) According to ypt-stage ypt1-2 84% (79 87) 78% (74 83) 86% (81 89) ypt3-4 60% (55 65) 52% (46 57) 63% (58 68) According to ypn-stage ypn0 81% (79 84) 76% (73 79) 81% (79 84) ypn1 52% (43 60) 47% (38 55) 59% (50 67) According to combined ypt- and ypn-stage ypt1-2n0 85% (80 88) 79% (74 83) 86% (81 90) ypt1-2n1 78% (59 89) 76% (58 88) 88% (70 95) ypt3-4n0 68% (62 74) 59% (52 64) 69% (63 75) ypt3-4n1 43% (32 53) 36% (26 46) 50% (38 60) cm 5 centimetre; LAR 5 low anterior resection; APR 5 abdominoperineal resection; pcr 5 pathologic complete response. plots. 32 Trend tests based on Schoenfeld residuals were also used to assess departure of proportionality. 32 Hazard ratios (HR) with 95% confidence intervals (CI) were calculated. The HRs that correspond with the variable coding for whether the patient received act or not indicate the relative risks of having an event during follow-up. A HR < 1 indicates that the patients who received act had a lower probability of having an event. To evaluate whether the effect of act was modified by the degree of response to CRT an interaction term for act and response to CRT was entered into the multivariable model. The three response groups were coded by two dummy variables using the response group with pcr as reference group. The HR associated with the interaction terms represent the ratio between the HR for response group ypt1-2 (or response group ypt3-4) and the HR of the reference group of patients with a pcr, indicating to what extent the effect of act is modified by the response to CRT. A similar analysis was performed with an interaction term for ypn-stage and act. Analyses were performed with SPSS 16.0 and STATA p-values < 0.05 were considered statistically significant. Results In total 3,313 patients were included, of whom 1,723 (52%) underwent act. act consisted of 5-FU with leucovorin in almost all patients. Some patients had additional other chemotherapeutic agents (e.g., oxaliplatin) together with 5-FU. CRT consisted of Gy in fractions combined with FUbased chemotherapy. There were 898 patients with a pcr, 966 with ypt1-2 tumours and 1,302 with ypt3-4 tumours.

7 218 Adjuvant chemotherapy in rectal cancer Figure 1. Forest plots displaying the HR with 95% confidence interval derived from the comparison of patients treated with adjuvant chemotherapy and not treated with adjuvant chemotherapy (reference group) for the three response groups. (a) RFS, (b) DFS and (c) OS. Factors in the multivariable Cox regression analyses were age, gender, ct-stage, cn-stage, type of surgery, ypn-stage and obviously administration of adjuvant chemotherapy. These analyses were stratified for the data providing center. Approximately half of the patients with pcr (441/898) came from the study by Capirci et al., accounting for the relatively high number of patients with pcr. 9 The majority of patients had a ypn0 status (n 5 2,591, 80%). Median follow-up for all patients was 51 (range 0 219) months. Additional information and baseline characteristics per study are shown in Table 1. Baseline characteristics for all patients and patients who did and did not undergo act are shown in Table 2. Patients who underwent act were significantly younger than patients who did not undergo act. Furthermore, patients who underwent act had more proximal tumors, underwent a low anterior resection more often and had more advanced pathologic stages (higher ypt- and ypn-stage) than patients who did not undergo act (all p < ). In Table 3, information on baseline characteristics are shown and compared per response category. Almost all patients had ct3 tumours before CRT (82% in the pcr group, 84% in the ypt0-2 group and 82% in the ypt3-4 group). In every of the three groups, approximately half of patients had involved nodes before CRT. Table 4 provides information on the prognostic value of ypt- and ypn- staging by showing the 5-year Kaplan Meier estimates of RFS, DFS and OS for patients not treated with act according to histological yptn categories. Multivariable analyses Multivariable Cox regression analyses were performed with inclusion of the following factors as independent variables: age, gender, clinical T-stage, clinical N-stage, type of surgery, administration of act. The proportional hazards assumption was not violated. When analyzing the combined data of all patients, the adjusted HRs indicated a better prognosis for patients who received act than for patients who did not. The adjusted HR for the effect of act on RFS was 0.81 (95% CI ), for DFS the HR was 0.75 (95% CI ) and for OS the HR was 0.69 (95% CI ). Figure 1 shows the forest plots displaying the HRs and 95% CIs for administration of act per ypt and per ypnstage for RFS, DFS and OS. For patients with a pcr, the HR vary around 1 with wide 95% CIs. For RFS, the HR was 1.25 (95% CI: ) for patients with pcr. For DFS and OS the HRs for this response group were 0.94 (95% CI: ) and 0.88 (95% CI: ), respectively. For patients with ypt1-2 and ypt3-4, the HRs for RFS were 0.58 ( ) and 0.83 ( ), respectively. For DFS and OS the HRs were all <1 showing an obvious benefit for administration of act in these patient groups. A similar effect was seen when analysing the data for local recurrence and distant metastasis separately (data not shown). For patients with ypn0, the HR for act on RFS was 0.74 (95% CI: ), while in patients with ypn1 the HR was 0.98 (95% CI: ). For DFS, similar results were found and for OS the difference in the HRs for act between patients with ypn0 and ypn1 was less pronounced: HR 0.67 (95% CI: ) for ypn0 and HR 0.74 (95% CI: ) for ypn1. To evaluate the robustness of the results for patients with pcr, we repeated the analysis without the patients with pcr from Capirci et al. 9 The resulting HR of 1.24 (95% CI: ) for RFS suggested no benefit from act for this subgroup. Interaction For RFS, the HRs for the interaction terms suggest a stronger effect of act in the response groups ypt1-2 (HR with 95% CI: , p ) and ypt3-4 (HR with 95% CI: , p ) than in the subgroup with pcr, but the estimated interaction effects do not reach statistical significance. This analysis was repeated with an interaction term for ypn status and act. For RFS, the HR for the interaction term was 0.80 (95% CI: , p ), indicating a larger effect of act in patients with ypn0 stage. For DFS and OS, the interaction HRs were also nonsignificant. Discussion This pooled analysis suggests that the histological staging after CRT and surgery for rectal cancer correlates with the effect of subsequent act. Patients with pcr appear to have an excellent prognosis regardless of use of act. The potential benefit of act in patients with a pcr after CRT would appear to be small if it exists at all; however, proving the absence of such a small benefit would require a large

8 Maas et al. 219 randomized prospective trial and may not be warranted. 10 The results for patients with residual tumour that receive act are compatible with an improvement of the long-term outcome, with the largest benefit for the patients with ypt1-2 tumours after CRT. In 2010, a review of randomized trials suggested that the effect of act in patients with rectal cancer treated with CRT was limited at best, and no single trial was large enough to detect a 5% improvement in 5-year survival. 5 In 2012, a metaanalysis of 9,785 patients with rectal cancer by the Cochrane collaboration found a significant improvement of DFS and OS with HRs of 0.75 and 0.83, respectively, when adjuvant 5-FU based chemotherapy was administered. It was pointed out that there is much uncertainty as to which patient groups (according to TNM stage) might benefit most, particularly in the setting of multimodality neoadjuvant treatment. The results of this present pooled analysis with a HR for RFS of 0.81 (95% CI ) are in accordance with a trend towards a survival benefit from act that was shown in two randomized trials that were included in the Cochrane review and used neoadjuvant treatment in a part of the patients. 33,34 Subgroup results of the QUASAR trial focusing on patients with rectal cancer showed a HR of 0.60 (95% CI: ), in favor of act. 33 Our finding that the effect of act is stronger in the ypt1-2 group than in the ypt3-4 group is consistent with the results from the subgroup analysis of the EORTC trial as published in 2007, where act was clearly beneficial for ypt0-2 tumours and no benefit was reported in patients with ypt A hypothesis is that the same factors drive both tumour sensitivity for the primary treatment with CRT and sensitivity for act aimed at microscopic distant disease. 8 The finding in the current pooled analysis that act seems to have an effect in patients with ypn0 tumours, but not in patients with ypn1 tumours also supports this hypothesis. However, a recent update from the EORTC trial with a median follow-up of 10 years did not confirm the earlier findings of the different effect or chemotherapy according to the response, and with again no overall benefit of adjuvant 5- FU systemic therapy. 35 One of the issues with the EORTC trial is the poor adherence to chemotherapy with <43% of patients receiving the planned dose in the predefined time. The authors suggested that this poor adherence to chemotherapy could explain the lack of effect of chemotherapy in the intention to treat analysis of the randomized trial, whereas the current pooled analysis is based on actual received treatment. Given these conflicting results the definitive effect of chemotherapy on long-term outcome in (subgroups of) rectal cancer remains unclear. In colon cancer (not treated with neoadjuvant therapy), the decision to administer act is guided by the pathologic N-stage: patients with pathologic N1 stage have an indication for act. However, given the findings in this pooled analysis, in patients with rectal cancer who have been treated with CRT and have a pathologic N1 status might not benefit from act. These results show that decision making about act should be approached differently in neoadjuvantly treated patients with rectal cancer than in patients with colon cancer. The results of this pooled analysis should be interpreted with caution because data were not derived from randomized trials. Due to confounding by indication the distribution of prognostic factors for survival may have differed between patients who received act and those who did not. An attempt was made to eliminate this bias by adjustment for these differences in a multivariable Cox regression model, but it is possible that some relevant prognostic factors were not captured in these multivariable analyses. For instance, no information was available on postoperative complications or comorbidity of patients. Both factors may have guided the decision to withhold act. As a result, the group of patients who did not receive act may have had an elevated risk of death due to other diseases than rectal cancer. Such an imbalance would result in bias towards favoring act for the outcomes DFS and OS. However, the HRs for RFS are less likely to be biased, because this outcome does not comprise deaths and can, therefore, be considered a more valid and thus reliable estimate of effectiveness of act. A second limitation is that the majority of patients received 5-FU-based act. For colon cancer it has been shown that adding oxaliplatin to 5-FU and leucovorin improves DFS for stage III patients and leads to a (nonsignificant benefit) of 5.4% for Stage II patients. 36 It remains unclear whether the results of the present study are reproducible with a regimen that includes oxaliplatin. A third limitation is that, although the results of the current pooled analysis support the presence of differential effects of act within different subgroups, there is no definite proof of treatment-by-response interaction. The HRs for the interaction terms suggest a stronger effect of act in the response groups ypt1-2 and ypt3-4 than in the subgroup with pcr, but the estimated interaction effects do not reach statistical significance. Tests for interaction are known to have limited power. Therefore, despite the large sample size, the power of this analysis to detect a significant interaction effect of this magnitude was only 39%. However, it cannot be concluded that lack of statistical significance indicates absence of interaction. The considerable differences in effect of act between the response groups strongly suggest the presence of effect modification by response to CRT even though the interaction test was not significant. Resolving the limitations that are inherent to pooled analyses of nonrandomized individual patient data, a randomized controlled trial (RCT) or a pooled analysis of data from RCTs, would be the best design. However, again, a very large sample size is needed to achieve enough power to detect small but clinically relevant improvements in survival. This is particularly challenging when trying to obtain a large enough sample size of patients with a pcr, as the prevalence is only around 15 20%. 10 Additionally, as chemotherapeutic regimens are often updated with newer (and possibly more effective) agents, the results of a RCT with the currently applied chemotherapeutic agents might no longer be clinically relevant upon completion of the RCT. Therefore, to provide definiteevidencewitharctwillbeverychallenging.while

9 220 Adjuvant chemotherapy in rectal cancer awaiting the results of such a RCT, the results of the current pooled analysis might therefore be used when contemplating on the use of act in individual patients. Conclusions act in patients treated with CRT for locally advanced rectal cancer results in improved prognosis and treatment effects differ between subgroups. Patients with a pcr after CRT may not benefit from act, whereas patients with residual tumour do. Although the test for interaction did not reach statistical significance, the results strongly support further investigation of a more individualized approach to administer act after CRT and surgery based on final pathologic staging. References 1. Jonker DJ, Spithoff K, Maroun J. Adjuvant systemic chemotherapy for stage II and III colon cancer after complete resection: an updated practice guideline. Clin Oncol (R Coll Radiol) 2011;23: Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. N Engl J Med 2004;351: Bipat S, Glas AS, Slors FJ, et al. Rectal cancer: local staging and assessment of lymph node involvement with endoluminal US, CT, and MR imaging a meta-analysis. Radiology 2004;232: Lahaye MJ, Engelen SM, Nelemans PJ, et al. Imaging for predicting the risk factors the circumferential resection margin and nodal disease of local recurrence in rectal cancer: a meta-analysis. Semin Ultrasound CT MR 2005;26: Bujko K, Glynne-Jones R, Bujko M. Does adjuvant fluoropyrimidine-based chemotherapy provide a benefit for patients with resected rectal cancer who have already received neoadjuvant radiochemotherapy? a systematic review of randomised trials. Ann Oncol 2010;21: Petersen SH, Harling H, Kirkeby LT, et al. Postoperative adjuvant chemotherapy in rectal cancer operated for cure. Cochrane Database Syst Rev 2012;3:CD Valentini V, van Stiphout RG, Lammering G, et al. Nomograms for predicting local recurrence, distant metastases, and overall survival for patients with locally advanced rectal cancer on the basis of European randomized clinical trials. J Clin Oncol 2011;29: Collette L, Bosset JF, den Dulk M, et al. Patients with curative resection of ct3-4 rectal cancer after preoperative radiotherapy or radiochemotherapy: does anybody benefit from adjuvant fluorouracil-based chemotherapy? a trial of the European Organisation for Research and Treatment of Cancer Radiation Oncology Group. J Clin Oncol 2007;25: Capirci C, Valentini V, Cionini L, et al. Prognostic value of pathologic complete response after neoadjuvant therapy in locally advanced rectal cancer: long-term analysis of 566 ypcr patients. Int J Radiat Oncol Biol Phys 2008;72: Maas M, Nelemans PJ, Valentini V, et al. Longterm outcome in patients with a pathological complete response after chemoradiation for rectal cancer: a pooled analysis of individual patient data. Lancet Oncol 2010;11: Goethals L, Haustermans K, Perneel C, et al. Chemo-radiotherapy versus radiotherapy alone in the pre-operative treatment of resectable rectal cancer. Eur J Surg Oncol 2005;31: Mohiuddin M, Hayne M, Regine WF, et al. Prognostic significance of postchemoradiation stage following preoperative chemotherapy and radiation for advanced/recurrent rectal cancers. Int J Radiat Oncol Biol Phys 2000;48: Shivnani AT, Small W, Jr., Stryker SJ, et al. Preoperative chemoradiation for rectal cancer: results of multimodality management and analysis of prognostic factors. Am J Surg 2007;193:389 93; discussion Guillem JG, Chessin DB, Cohen AM, et al. Longterm oncologic outcome following preoperative combined modality therapy and total mesorectal excision of locally advanced rectal cancer. Ann Surg 2005;241: Govindarajan A, Reidy D, Weiser MR, et al. Recurrence Rates and Prognostic Factors in ypn0 Rectal Cancer After Neoadjuvant Chemoradiation and Total Mesorectal Excision. Ann Surg Oncol 2011;18: Biondo S, Navarro M, Marti Rague J, et al. Response to neoadjuvant therapy for rectal cancer: influence on long-term results. Colorectal Dis 2005;7: Diaz-Gonzalez JA, Calvo FA, Cortes J, et al. Prognostic factors for disease-free survival in patients with T3-4 or N1 rectal cancer treated with preoperative chemoradiation therapy, surgery, and intraoperative irradiation. Int J Radiat Oncol Biol Phys 2006;64: Garcia-Aguilar J, Hernandez de Anda E, Sirivongs P, et al. A pathologic complete response to preoperative chemoradiation is associated with lower local recurrence and improved survival in rectal cancer patients treated by mesorectal excision. Dis Colon Rectum 2003;46: Hughes R, Glynne Jones R, Grainger J, et al. Can pathological complete response in the primary tumour following pre-operative pelvic chemoradiotherapy for T3-T4 rectal cancer predict for sterilisation of pelvic lymph nodes, a low risk of local recurrence and the appropriateness of local excision? Int J Colorectal Dis 2006;21: Janjan NA, Crane C, Feig BW, et al. Improved overall survival among responders to preoperative chemoradiation for locally advanced rectal cancer. Am J Clin Oncol 2001;24: Kuo LJ, Liu MC, Jian JJ, et al. Is final TNM staging a predictor for survival in locally advanced rectal cancer after preoperative chemoradiation therapy? Ann Surg Oncol 2007;14: Pucciarelli S, Toppan P, Friso ML, et al. Complete pathologic response following preoperative chemoradiation therapy for middle to lower rectal cancer is not a prognostic factor for a better outcome. Dis Colon Rectum 2004;47: Rodel C, Martus P, Papadoupolos T, et al. Prognostic significance of tumor regression after preoperative chemoradiotherapy for rectal cancer. J Clin Oncol 2005;23: Suarez J, Vera R, Balen E, et al. Pathologic response assessed by Mandard grade is a better prognostic factor than down staging for diseasefree survival after preoperative radiochemotherapy for advanced rectal cancer. Colorectal Dis 2008;10: Theodoropoulos G, Wise WE, Padmanabhan A, et al. T-level downstaging and complete pathologic response after preoperative chemoradiation for advanced rectal cancer result in decreased recurrence and improved disease-free survival. Dis Colon Rectum 2002;45: Valentini V, Coco C, Picciocchi A, et al. Does downstaging predict improved outcome after preoperative chemoradiation for extraperitoneal locally advanced rectal cancer? A long-term analysis of 165 patients. Int J Radiat Oncol Biol Phys 2002;53: Hughes R, Glynne-Jones R, Grainger J, et al. Can pathological complete response in the primary tumour following pre-operative pelvic chemoradiotherapy for T3-T4 rectal cancer predict for sterilisation of pelvic lymph nodes, a low risk of local recurrence and the appropriateness of local excision? Int J Colorectal Dis 2006;21: Biondo S, Navarro M, Marti-Rague J, et al. Response to neoadjuvant therapy for rectal cancer: influence on long-term results. Colorectal Dis 2005;7: Valentini V, Coco C, Picciocchi A, et al. Does downstaging predict improved outcome after preoperative chemoradiation for extraperitoneal locally advanced rectal cancer? A long-term analysis of 165 patients. Int J Radiat Oncol Biol Phys 2002;53: Govindarajan A, Reidy D, Weiser MR, et al. Recurrence rates and prognostic factors in ypn0 rectal cancer after neoadjuvant chemoradiation and total mesorectal excision. Ann Surg Oncol 2011;18: Guillem JG, Chessin DB, Cohen AM, et al. Longterm oncologic outcome following preoperative combined modality therapy and total mesorectal excision of locally advanced rectal cancer. Ann Surg 2005;241:829 36; discussion Bellera CA, MacGrogan G, Debled M, et al. Variables with time-varying effects and the Cox model: some statistical concepts illustrated with a prognostic factor study in breast cancer. BMC Med Res Methodol 2010;10: Gray R, Barnwell J, McConkey C, et al. Adjuvant chemotherapy versus observation in patients with colorectal cancer: a randomised study. Lancet 2007;370: Bosset JF, Collette L, Calais G, et al. Chemotherapy with preoperative radiotherapy in rectal cancer. N Engl J Med 2006;355: Bosset JF, Calais G, Mineur L, et al. Fluorouracilbased adjuvant chemotherapy after preoperative chemoradiotherapy in rectal cancer: long-term results of the EORTC randomised study. Lancet Oncol 2014;15: Wolpin BM, Meyerhardt JA, Mamon HJ, Mayer RJ. Adjuvant treatment of colorectal cancer. CA Cancer J Clin 2007;57: