Neoadjuvant chemoradiotherapy for esophageal carcinoma: A meta-analysis

Neoadjuvant chemoradiotherapy for esophageal carcinoma: A meta-analysis Sarah E. Greer, MD,a Philip P. Goodney, MD,a,b John E. Sutton, MD,a and John D. Birkmeyer, MD,a,b Lebanon, NH, and White River Junction, Vt

Background. The effectiveness in improving survival of neoadjuvant chemoradiotherapy (NCRT) in patients undergoing surgery for esophageal carcinoma remains unclear. Methods. MEDLINE, the Cochrane Database of Systematic Reviews, BIOSIS Previews, and other resources were searched from January 1966 through January 2003. Randomized trials were selected on the basis of study design (NCRT followed by surgery vs surgery alone). Of 21 potential studies identified by abstract review, 6 (29%) met the inclusion criteria. Results. Across 6 studies, a total of 374 patients underwent NCRT followed by surgery and 364 underwent surgery alone. In 5 of the 6 studies in our meta-analysis, there was a small, non---statistically significant trend toward improved survival with NCRT. Only 1 study demonstrated a statistically significant benefit to NCRT. In our summary measure for all 6 studies, we found a small, non--statistically significant trend toward improved long-term survival in the NCRT followed by surgery group (relative risk of death in the NCRT group [RR], 0.86; 95% confidence interval [CI], 0.74 to 1.01; P = .07). Conclusions. NCRT followed by surgery is associated with a small, non--statistically significant improvement in overall survival. Whether this benefit is sufficient to warrant the considerable expense and risks associated with NCRT should be the subject of future larger randomized trials. (Surgery 2005;137:172-7.) From the Department of Surgery, Dartmouth-Hitchcock Medical Center,a Lebanon, and the VA Outcomes Group, Department of Veterans Affairs Medical Center,b White River Junction

DESPITE ADVANCES IN SURGICAL TECHNIQUE, overall 5year survival rates for esophageal cancer remain low, averaging below 30%.1-5 Although surgery remains the standard treatment, there is growing support for neoadjuvant chemoradiotherapy (NCRT). Several groups6-10 have combined neoadjuvant radiotherapy (typically ranging from 30-60 Gy) with neoadjuvant cisplatin-based chemotherapy regimens in an attempt to increase survival by ‘‘downstaging’’ patients11 before resection or even eradicating the detectable tumor burden.6,12 Several studies6,9,13 have also suggested improved long-term survival rates with NCRT. For example, in one case series survival improved from 17%14 to 34%7 after the

Accepted for publication June 25, 2004. Reprint requests: Sarah E. Greer, MD, Dartmouth-Hitchcock Medical Center, Lebanon, NH, 03756. E-mail: sarah.greer@ hitchcock.org. The views expressed herein do not necessarily represent the views of the Department of Veterans Affairs or the United States Government. 0039-6060/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.surg.2004.06.033

172 SURGERY

implementation of a regimen of preoperative radiotherapy, cisplatin, fluorouracil, and vinblastine. However, whether or not NCRTactually increases long-term survival remains controversial. First, many studies reporting beneficial effects with NCRT have been observational in design. As with all observational studies, the absence of control groups in these studies makes it difficult to evaluate the validity and strength of their findings. Second, the few randomized trials that have examined the effect of NCRT have been small,15-20 thus limiting their power to detect small differences. Many reviews of treatment options for esophageal cancer have been published;21-25 however, only one previous meta-analysis has been performed examining NCRT in esophageal cancer.26 Because the overall effect of NCRT on long-term survival remains uncertain, we used meta-analysis to evaluate the effect of NCRT on overall survival in patients undergoing surgery for esophageal cancer. METHODS Data sources. Data sources included studies that contained original data examining the effect of neoadjuvant therapy on overall survival in esophageal cancer. The MEDLINE database was searched

Greer et al 173

Surgery Volume 137, Number 2

from January 1966 through January 2003, using the terms ‘‘esophageal neoplasm,’’ ‘‘carcinoma,’’ ‘‘adjuvant,’’ ‘‘chemotherapy,’’ ‘‘radiotherapy,’’ and ‘‘combined modality therapy’’ (Fig 1). Further articles were identified from the reference sections of relevant studies, letters, editorials, comments, and books. This strategy was supplemented by using the Cochrane database of systematic reviews and biosis previews to generate a list of articles for subsequent review. Study selection. Only randomized controlled trials that directly compared NCRT followed by surgery with surgery alone were included. We identified overall survival as our main outcome measure because this metric was the one most commonly identified across studies. Two investigators (S.E.G. and P.P.G.) reviewed potential citations and selected appropriate studies. After the initial MEDLINE search, we retrieved 21 potential studies containing research relevant to NCRT. Of these studies, 6 met our inclusion criteria and were included in the meta-analysis. Analysis. One investigator (S.E.G.) extracted the following data for each study: design, type of esophageal cancer (squamous or adenocarcinoma), staging methods, NCRT regimen, type of surgery, outcomes measured, and mean length of follow-up. Ideally, individual patient data would be used in meta-analyses, which would allow more direct comparisons and help eliminate bias. In meta-analysis of time-to-event analyses, information about follow-up periods for each patient is especially important for weighting observations.27 However, despite multiple attempts, we were unsuccessful in obtaining this information from authors of the 6 randomized controlled trials.15-20 For this reason, we were required to weight studies according to their mean follow-up periods as reported by these studies. When means were not reported, we estimated means from reported medians using the following formula: mean = (ÿ1/median)(ln 0.5).28,29 If neither mean nor median length of follow-up could be obtained, then we assumed the length of follow-up to be equal to one half the duration of overall survival reported in the study. In addition, we divided the number of deaths by the patient-months of follow-up within each arm of each study, to allow comparison of studies with different lengths of follow-up. We then used standard meta-analytic techniques to calculate our summary statistics. We used a random-effects model to calculate a summary value for relative risk.30 Before pooling the results of these different studies, we determined that the data were not significantly different by testing for homogene-

Fig 1. Search strategy.

ity across studies.31 We performed our analyses using Review Manager version 4.1 (Cochrane Collaboration; the Nordic Cochrane Centre, Copenhagen, Denmark). P values <.05 were considered statistically significant. RESULTS Study characteristics. We identified 6 studies that met our inclusion criteria (Table I).15-20 Across these studies, a total of 374 patients were treated with NCRT followed by surgery and 364 patients were treated with surgery alone. Only 2 of the 6 studies19,20 included patients with adenocarcinoma; most focused on squamous cell carcinoma. Although nearly all patients in the surgery-alone arm completed surgery, only 72% to 97% of patients in the NCRT followed by the surgery arm were able to complete both NCRT and subsequent surgery. Most patients were stage I or stage II at the time of study enrollment. However, 2 studies15,20 enrolled patients with more advanced disease. Stage at the time of diagnosis by study treatment arm was unavailable in one case.19 In another case, only the stage of disease post-NCRT was reported.20 Neoadjuvant modalities included cisplatin-based chemotherapy regimens in all cases (Table II), in combination with fluorouracil in 4 of the 6 trials.15,17,19,20 Other chemotherapeutic agents used included bleomycin and vinblastine. Total

174 Greer et al

Surgery February 2005

Table I. Characteristics of patients in the 6 studies in our analysis Patients stage III or greater, no. (percent)

Subjects (n) Author and year Apinop et al, 1994 Bossett et al, 1997 LePrise et al, 1994 Nygaard et al, 1992 Urba et al, 2001 Walsh et al, 1996 Totals

Study type

NCRT

Surgery alone

35 143 41 47 50 58 374

34 139 45 41 50 55 364

RCT RCT RCT RCT RCT RCT 6 RCTs

Type of cancer

Stages enrolled

Squamous Squamous Squamous Squamous Both Adenocarcinoma 4 squamous; 1 adenocarcinoma; 1 both

IIb-III I-IIb I-II I-IIb I-III* 0-IVy Range, 0-IV

NCRT 29 0 0 0 – 16 45

(83%) (0%) (0%) (0%) (28%) (14%)

Surgery alone 28 0 0 0 – 43 71

(82%) (0%) (0%) (0%) (78%) (23%)

P < .006 *Stage extrapolated from reported tumor size. yStaging done post-NCRT.

Table II. Treatment characteristics of each of the studies in our analysis Author and year

Chemotherapy regimen

Dose of radiotherapy

Type of resection

Outcomes Overall survival Overall survival, disease-free survival Overall survival, disease-free survival Overall survival Overall survival, disease-free survival Overall survival

Apinop et al, 1994 Bossett et al, 1997

CDDP,* 5-FUy CCDP

40 Gy 37 Gy

Thoracoabdominal Thoracoabdominal

LePrise et al, 1994

CDDP, 5-FU

20 Gy

Not stated

Nygaard et al, 1992 Urba et al, 2001

CDDP, bleomycin CDDP, 5-FU

35 Gy 45 Gy

Thoracoabdominal Transhiatal

Walsh et al, 1996

CDDP, 5-FU

40 Gy

Both

*CDDP, cis-diamminedichloroplatinum (cisplatin). y5-FU, 5-flourouracil.

doses of radiotherapy varied between 20 and 45 Gy and were administered over 2 to 4 weeks. Surgical intervention varied across studies. Three of the 6 studies used thoracoabdominal resections to perform esophagectomy. Of the remaining studies, 1 study used transhiatal resection,19 1 study used a combination of thoracoabdominal resection and transhiatal resection,20 and 1 study did not describe the operative technique.17 Overall survival. All 6 trials reported overall survival. Three of the studies reported 5-year survival,15,16,19 and 3 reported 3-year survival.17,18,20 Five of the 6 studies reported small, non–statistically significant improvements in long-term survival in patients undergoing NCRT. Only 1 study reported a statistically significant improvement in survival in patients undergoing NCRT. Our summary measure of survival indicated that the relative risk (RR) of death was 0.86 for those patients treated with NCRT followed by surgery compared with surgery alone, but this effect was not statistically significant (95% CI, 0.74 to 1.01; P = .07) (Fig 2).

One factor that may have contributed to the lack of effect in the NCRT arm was death before surgery. In our analysis, 4 deaths before surgery occurred in the NCRT group. These 4 deaths were attributed to hemorrhage from the tumor bed during treatment, esophageal perforation, electrolyte imbalance, and ‘‘toxic effects.’’ Deaths before surgery occurred in 3 studies15,16,20 with regimens of cisplatin alone or cisplatin and fluorouracil, and total doses of 37 to 40 Gy of radiation. Finally, to examine the effect of different types of esophageal cancer, we limited our analysis to only the 4 studies that included patients with squamous cell carcinoma. NCRT in this subgroup was associated with a relative risk of death was 0.94 (95% CI, 0.77 to 1.13, P = .5), not significantly different than the results from our combined analysis. DISCUSSION We performed a meta-analysis of 6 randomized controlled trials comparing NCRT followed by surgery versus surgery alone for treating esophageal

Surgery Volume 137, Number 2

Greer et al 175

Fig 2. Relative risk (RR) of death for NCRT followed by surgery compared with surgery alone, measured by number of deaths (n)/patient-months of follow-up (N).

cancer. In 5 of the 6 studies in our meta-analysis there was a small, non--statistically significant trend toward improved survival with NCRT. The remaining study showed a small, statistically significant benefit of NCRT. Our summary measure, incorporating data from 6 studies and more than 700 patients, revealed a small, non--statistically significant survival benefit with NCRT. This study has several limitations. First, although patients were similar across the studies in many ways, some differences are worth noting. Combining patients with adenocarcinoma and squamous cell carcinoma of the esophagus may be disputed. However, given the rare nature of esophageal cancer, a similarly poor prognosis for both of these histologies, and similar chemotherapy and radiotherapy treatment regimens, we felt that it was reasonable to group these patients in our analysis. In fact, Urba et al19 included both adenocarcinoma and squamous cell carcinoma in their trial. Additionally, our subgroup analysis limited to squamous cell carcinoma resulted in a relative risk of death similar to our main finding, further supporting our decision. Second, there were significantly more patients with advanced disease (stage III or greater) in the surgery-alone arm. However, only 2 studies15,20 enrolled patients with advanced disease, and Walsh20 reported only the stage of disease postNCRT. This may have resulted in an overall ‘‘downstaging’’ of patients in the NCRT group and a false impression that patients in the surgery-alone arm had more advanced disease. Uncertainty about the true baseline characteristics of patients limits our ability to interpret the effect of preoperative stage on outcome. Although therapeutic regimens were cisplatinbased in all 6 studies, there were some differences in the chemotherapy agents used and radiotherapy doses given. It is difficult to determine whether differing results with respect to overall survival are

due to more effective chemoradiotherapy regimens in some trials. Surgical technique was not uniform across the 6 studies. Although the debate over transthoracic versus transhiatal resection continues, there has been evidence to support the equality of these 2 techniques with respect to outcomes.32,33 Also, despite the confines of clinical trials, between-surgeon variability in operative strategies exists and is likely to continue. Meta-analyses of long-term survival in cancer trials are optimally accomplished by interpreting patient-level data, which allows more direct comparisons and helps eliminate bias. Unfortunately, as with most published meta-analyses, patient-level data were not available for our study. To determine length of follow-up, we used mean values from each arm of each study rather than values from individual patients. Interpreting follow-up at this level could introduce bias, especially if the patients in the NCRT arm survived longer. However, this bias would tend to cause our summary measure to slightly underestimate the true effect. Moreover, our summary measure indicated a small, nonsignificant improvement in survival, an outcome replicated in 5 of the 6 studies in our analysis. Such close agreement between the outcomes of the studies and the summary measure makes it unlikely that significant bias was an important factor in our analysis. To date, only one earlier meta-analysis examining the effectiveness of NCRT in patients with esophageal cancer has been published. Kaklamanos et al26 assessed neoadjuvant chemotherapy both alone and in combination with preoperative radiotherapy. Their summary measure showed a non–statistically significant 6.4% absolute improvement in 2-year survival (95% CI, ÿ1.2 to 14.0; P = .86) associated with NCRT. Although the conclusions of that study mirror our conclusions overall, that study focused on 2-year survival. In contrast, our analysis

176 Greer et al

incorporated later survival data (up to 5 years) from these studies. Moreover, the study by Kaklamanos et al26 included only 5 of the 6 trials examined in our study. For both of these reasons, our findings may be more robust. In conclusion, although NCRT is currently used in many settings, evidence to support its routine administration is lacking. Although some encouraging data support using NCRT to treat esophageal cancer, significant uncertainty about the true longterm survival benefits of NCRT remains. Although our analysis suggested a trend toward improved survival, this benefit was relatively small, and its clinical significance may be best judged by an individual well-informed patient. It is important to note that longevity is not the only consideration in making decisions about NCRT in esophageal cancer. For this reason, future clinical trials should focus on other important outcomes, including quality of life, morbidity, and cost.

REFERENCES 1. SEER cancer statistics review, 1973-1999. Bethesda, MD: National Cancer Institute, 2002; available at http://seer. cancer.gov/csr/1973_1999. 2. Parker SL, Tong T, Bolden S, Wingo PA. Cancer statistics, 1997. CA Cancer J Clin 1997;47:5-27. 3. Wingo PA, Ries LA, Parker SL, Heath CW Jr. Long-term cancer patient survival in the United States. Cancer Epidemiol Biomarkers Prev 1998;7:271-82. 4. Farrow DC, Vaughan TL. Determinants of survival following the diagnosis of esophageal adenocarcinoma (United States). Cancer Causes Control 1996;7:322-7. 5. Thomas RM, Sobin LH. Gastrointestinal cancer. Cancer 1995;75(Suppl 1):154-70. 6. Chidel MA, Rice TW, Adelstein DJ, Kupelian PA, Suh JH, Becker M. Resectable esophageal carcinoma: local control with neoadjuvant chemotherapy and radiation therapy. Radiology 1999;213:67-72. 7. Forastiere AA, Orringer MB, Perez-Tamayo C, Urba SG, Zahurak M. Preoperative chemoradiation followed by transhiatal esophagectomy for carcinoma of the esophagus: final report. J Clin Oncol 1993;11:1118-23. 8. Poplin E, Fleming T, Leichman L, Seydel HG, Steiger Z, Taylor S, et al. Combined therapies for squamous cell carcinoma of the esophagus: a Southwest Oncology Group Study (SWOG-8037). J Clin Oncol 1987;5:622-8. 9. MacFarlane SD, Hill LD, Jolly PC, Kozarek RA, Anderson RP. Improved results of surgical treatment for esophageal and gastroesophageal junction carcinomas after preoperative combined chemotherapy and radiation. J Thorac Cardiovasc Surg 1988;95:415-22. 10. Seydel HG, Leichman L, Byhardt R, Cooper J, Herskovic A, Libnock J, et al. Preoperative radiation and chemotherapy for localized squamous cell carcinoma of the esophagus: a RTOG study. Int J Radiat Oncol Biol Phys 1988;14: 33-5. 11. Bessell JR, Devitt PG, Gill PG, Goyal S, Jamieson GG. Prolonged survival follows resection of oesophageal SCC

Surgery February 2005

12.

13.

14.

15.

16.

17.

18.

19.

20.

21. 22. 23.

24.

25. 26.

27.

28.

downstaged by prior chemoradiotherapy. Aust NZ J Surg 1996;66:214-7. Stahl M, Wilke H, Fink U, Stuschke M, Walz MK, Siewert JR, et al. Combined preoperative chemotherapy and radiotherapy in patients with locally advanced esophageal cancer: interim analysis of a phase II trial. J Clin Oncol 1996;14: 829-37. Wolfe WG, Vaughn AL, Seigler HF, Hathorn JW, Leopold KA, Duhaylongsod FG. Survival of patients with carcinoma of the esophagus treated with combined-modality therapy. J Thorac Cardiovasc Surg 1993;105:749-55. Orringer MB. Transhiatal esophagectomy without thoracotomy for carcinoma of the thoracic esophagus. Ann Surg 1984;200:282-8. Apinop C, Puttisak P, Preecha N. A prospective study of combined therapy in esophageal cancer. Hepatogastroenterology 1994;41:391-3. Bosset JF, Gignoux M, Triboulet JP, Tiret E, Mantion G, Elias D, et al. Chemoradiotherapy followed by surgery compared with surgery alone in squamous cell cancer of the esophagus. N Engl J Med 1997;337:161-7. Le Prise E, Etienne PL, Meunier B, Maddern G, Ben Hassel M, Gedouin D, et al. A randomized study of chemotherapy, radiation therapy, and surgery versus surgery for localized squamous cell carcinoma of the esophagus. Cancer 1994;73: 1779-84. Nygaard K, Hagen S, Hansen HS, Hatlevoll R, Hultborn R, Jakobsen A, et al. Pre-operative radiotherapy prolongs survival in operable esophageal carcinoma: a randomized, multicenter study of pre-operative radiotherapy and chemotherapy. The second Scandinavian trial in esophageal cancer. World J Surg 1992;16:1104-9. Urba SG, Orringer MB, Turrisi A, Iannettoni M, Forastiere A, Strawderman M. Randomized trial of preoperative chemoradiation versus surgery alone in patients with locoregional esophageal carcinoma. J Clin Oncol 2001;19: 305-13. 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;335:462-7. DeMeester TR. Esophageal carcinoma: current controversies. Semin Surg Oncol 1997;13:217-33. Enzinger PC, Mayer RJ. Esophageal cancer. N Engl J Med 2003;349:2241-52. DeCamp MM Jr, Swanson SJ, Jaklitsch MT. Esophagectomy after induction chemoradiation. Chest 1999;116(Suppl 6): 466S-9S. Blom D, Peters JH, DeMeester TR. Controversies in the current therapy of carcinoma of the esophagus. J Am Coll Surg 2002;195:241-50. Lehnert T. Multimodal therapy for squamous carcinoma of the oesophagus. Br J Surg 1999;86:727-39. Kaklamanos IG, Walker GR, Ferry K, Franceschi D, Livingstone AS. Neoadjuvant treatment for resectable cancer of the esophagus and the gastroesophageal junction: a metaanalysis of randomized clinical trials. Ann Surg Oncol 2003; 10:754-61. Duchateau L, Pignon JP, Bijnens L, Bertin S, Bourhis J, Sylvester R. Individual patient versus literature-based metaanalysis of survival data: time to event and event rate at a particular time can make a difference, an example based on head and neck cancer. Control Clin Trials 2001;22:538-47. Cox DR, Oakes D. Analysis of survival data. New York: Chapman & Hall; 1984.

Surgery Volume 137, Number 2

29. Gobbi PG, Cavalli C, Federico M, Bertoloni D, Di Prisco UA, Rossi A, et al. Hodgkin’s disease prognosis: a directly predictive equation. Lancet 1988;8587:675-9. 30. Hedges LV, Olkin I. Statistical methods for meta-analysis. New York: Academic Press; 1985. 31. Daniel WW. Biostatistics: A foundation for analysis in the health sciences. 7th ed. New York: Wiley; 1999.

Greer et al 177

32. Hulscher JB, van Sandick JW, de Boer AG, Wijnhoven BP, Tijssen JG, Fockens P, et al. Extended transthoracic resection compared with limited transhiatal resection for adenocarcinoma of the esophagus. N Engl J Med 2002;347:1662-9. 33. Bolton JS, Teng S. Transthoracic or transhiatal esophagectomy for cancer of the esophagus: does it matter? Surg Oncol Clin N Am 2002;11:365-75.