Definitive Chemoradiation for Rectal Cancer: Is There a Role for Dose Escalation? A National Cancer Database Study

Abstract

BACKGROUND:

Surgery remains the standard of care in rectal cancer. Select patients will not undergo surgery for reasons such as medical inoperability or a watch-and-wait approach and instead are managed with definitive chemoradiation.

OBJECTIVE:

We used the National Cancer Database to identify overall survival and predictors thereof in the nonoperative management of patients with rectal cancer.

DESIGN:

This was a retrospective review.

SETTINGS:

This study used deidentified data from the National Cancer Database.

PATIENTS:

We queried the national cancer database from 2004 to 2014 for stage 1 to 3 rectal adenocarcinoma treated with only chemotherapy and radiation to definitive doses. Dose escalated therapy was defined as >54 Gy.

MAIN OUTCOME MEASURES:

Univariable and multivariable analyses were performed to identify sociodemographic, treatment, and tumor characteristics predictive of dose escalation and overall survival. Propensity-adjusted Cox proportional hazard ratios for survival were used to account for indication bias.

RESULTS:

Among the 6311 patients eligible for the study, 11% were treated with doses >54 Gy. Earlier stage and increased age/comorbidity patients were more likely to receive dose escalation, and patients with more recent treatment and treatment at an academic facility were less likely. The median follow-up time was 31 months (range, 2–154 mo). Three- and 5-year overall survival rates for all patients were 60% and 46%. Patients treated with dose escalation had a median survival of 33 months compared with 56 months for those treated with ≤54 Gy (p < 0.0001).

LIMITATIONS:

The main limitation is the inherent selection bias present in National Cancer Database studies. Important treatment details and outcomes as they relate to a definitive chemoradiation approach in rectal cancer are lacking. Salvage therapy was also not recorded, which in this population could be surgery.

CONCLUSIONS:

In this analysis, dose escalation in the nonoperative management of rectal cancer was associated with a lower overall survival compared with more conventional doses. Careful patient selection and enrollment on appropriate clinical trials may be warranted in the nonoperative setting. See Video Abstract at http://links.lww.com/DCR/B15.

Rectal cancer remains a commonly diagnosed malignancy in the United States, with >40,000 new cases each year.¹ For those patients who present with locally advanced disease, a multidisciplinary approach involving colorectal surgery, radiation oncology, and medical oncology is used. With the emergence of long-term data from multiple clinical trials, chemoradiotherapy is preferentially delivered neoadjuvantly for patients with T3/4 or node-positive disease followed by total mesorectal excision.^2–4 Published research looking at population-based data from the National Cancer Database (NCDB) has shown a steady increase in the use of neoadjuvant chemoradiation therapy before surgery with excellent outcomes.⁵

More recently, some institutions have begun to look at the possibility of using a watch-and-wait approach after completion of standard and dose-escalated chemoradiation.^6,7 This approach affords patients the opportunity to potentially avoid major surgery, possible permanent colostomy, and the change in quality of life associated with those events. The majority of the literature on this approach comes from Brazil, where an initial 70 patients with T2–4N0–2 distal rectal cancer were treated to a higher radiation dose (54 Gy) with 6 cycles of 5-fluorouracil every 3 weeks.⁸ Patients then had tumor response assessment at 8 to 10 weeks postradiation, with either continued surveillance or operative management. The results were quite impressive compared with the numbers cited above, with 50% of patients not requiring any surgical intervention. This same group updated their experience in terms of salvage, noting that successful salvage was still possible in >90% of patients treated in that manner, with an 80% rate of overall organ preservation.⁹ Given their high success rates, the interest in a nonoperative approach is increasing for both patients and practice providers, with ongoing protocols currently evaluating outcomes. In the present study we sought to use the NCDB to look at the use of a nonsurgical approach in patients with nonmeta-static rectal cancer in the United States and to see if there was any difference in outcome based on dose of radiation delivered.

PATIENTS AND METHODS

We conducted a retrospective review using deidentified data from the NCDB, which is exempt from institutional review board oversight. The NCDB is a tumor registry jointly maintained by the American Cancer Society and the American College of Surgeons for >1500 hospitals accredited by the Commission on Cancer across the United States. It is estimated that the database captures up to an estimated 70% of newly diagnosed malignancies each. We queried the database for patients with American Joint Committee on Cancer clinical stage 1 to 3 rectal adenocarcinoma (International Classification of Diseases for Oncology, Third Revision, histology codes 8140, 8210, 8260‐63, 8470, 8480, and 8481) diagnosed between 2004 and 2014. Figure 1 is a Consolidated Standards for Reporting of Trials diagram outlining the cohort selection criteria. We excluded patients who had surgery as part of their initial treatment management. We also excluded any patients with stage IV disease, no radiation therapy, or radiation directed at nonpelvic sites (ie, excluded any cases that were palliative). Also excluded were patients not treated with systemic therapy or those patients with <2 months of follow-up to account for immortal time bias. We defined a nondefinitive dose of radiation as <44 Gy and >73 Gy and excluded those patients from the cohort.

FIGURE 1.

CONSORT diagram showing patient selection criteria. XRT = radiotherapy; chemo = chemotherapy; F/u = follow up; chemoRT = chemoradiation; NR = not recorded; CONSORT = Consolidated Standards for Reporting of Trials.

Race was documented as white, black, or other. Comorbidity was quantified using the relatively well known Charlson–Deyo comorbidity index, which has been defined elsewhere.¹⁰ Stage was defined according to the seventh edition of the American Joint Committee on Cancer’s clinical group. Socioeconomic data in the patients’ residence census tract were provided as quartiles of the percentage of persons with less than a high school education and median household income. The facility type was assigned and defined according to the Commission on Cancer accreditation category. Locations were assigned based on data provided by the US Department of Agriculture Economic Research Service. Insurance status is documented in the NCDB as it appears on the patient’s hospital chart admission page. As is the case with all NCDB studies, the data used in the study are derived from a deidentified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology used or the conclusions drawn from these data by the investigator.

Data were analyzed using Medcalc version 18 (Ostend, Belgium). Summary statistics are presented for discrete variables. χ² tests compared sociodemographic, treatment, and tumor characteristics between the treatment groups. Overall survival was calculated in months from time of diagnosis to date of last contact or death. Kaplan–Meier curves were used to calculate cumulative probability of survival.¹¹ Log-rank statistics were used to test whether there was a statistically significant difference in the cumulative proportions across groups. A Cox proportional hazards model was used for multivariable survival analysis.¹² Because of the large nature of the data set, factors significant on univariable analysis were entered using a stepwise backward elimination process. Adjusted HRs and 95% CIs are reported, using an α level of 0.05 to indicate statistical significance.

Propensity score–matched survival analysis was used to account for indication bias attributed to lack of randomization between patients receiving conventional or escalated radiation doses.¹³ Multivariable logistic regression was used to calculate a propensity score indicative of conditional probability of receiving conventional or escalated radiation doses. The propensity model included observable variables associated with treatment selection on multivariable logistic regression. A Cox proportional hazards model was then constructed incorporating the propensity score but also excluding factors included in the propensity score calculation to avoid overcorrection. The assumption of balance was additionally validated by stratifying the data into propensity score–based quintiles and confirming that the difference in propensity score mean per quintile was <0.10.

RESULTS

Baseline patient characteristics are outlined in Table 1. The vast majority of patients (88%) were stage 2 or 3. The median total radiation dose was 50.4 Gy (range, 45.0–72.0 Gy) in 28 fractions (interquartile range, 28–33 fractions). Dose escalation was defined as any dose >54 Gy. Radiation was started at a median 40 days after diagnosis (interquartile range, 27–91 d). Chemotherapy was started a median 38 days after diagnosis (interquartile range, 25–55 d). The odds of receiving dose-escalated therapy increased with higher comorbidity score, age >65 years, and rural location. The odds of receiving dose-escalated therapy decreased with treatment at an academic facility, increased overall stage, increased T stage, private insurance, and more recent year of treatment (Table 2).

Table 1.

Patient demographics and clinical characteristics at baseline (n = 7131)

Characteristics	No. (%)
Sex
Men	4570 (64)
Women	2561 (36)
Race
White	5878 (82)
Black	826 (11)
Other	427 (7)
Comorbidity score
0	5707 (80)
1	1050 (15)
≥2	374 (5)
Insurance
Not insured	390 (5)
Private payer	2277 (32)
Government	4347 (61)
Unrecorded	117 (2)
Education %
≥29.0	1495 (21)
20.0 to 28.9	1918 (27)
14.0 to 19.9	2195 (31)
<14.0	1438 (20)
Unrecorded	85 (1)
Treatment facility type
Community cancer program	962 (13)
Comprehensive community cancer program	3084 (43)
Academic/research program	2880 (40)
Unrecorded	206 (4)
Treatment facility location
Metro	5663 (80)
Urban	1081 (15)
Rural	163 (2)
Unrecorded	224 (3)
Income, US $
<30,000	1470 (21)
30,000 to 35,000	1663 (23)
35,000 to 45,999	1874 (26)
>46,000	2034 (29)
Unrecorded	90 (1)
Distance to treatment facility, miles
≤8.5	3517 (50)
>8.5	3614 (50)
Age distribution, y
≤65	3383 (50)
>65	3748 (50)
Year of diagnosis
2004–06	1327 (19)
2007–09	1790 (25)
2010–12	2101 (29)
2013–14	1913 (26)
Stage grouping
1	836 (12)
2	3316 (47)
3	2979 (41)

Table 2.

Comparative use of escalated radiation dose (defined as >54 Gy) by baseline characteristics in patients receiving concurrent chemoradiotherapy without surgery for rectal cancer

Characteristic	Conventional dose (n = 6311), n (%)	Escalated radiation dose (n = 820), n (%)	OR	95% CI	p
Sex
Men	4058 (64)	512 (62)	1	Reference
Women	2253 (36)	308 (38)	1.09	0.93–1.26	0.30
Race
White	5191 (82)	687 (84)	1	Reference
Black	728 (12)	98 (12)	1.02	0.81–1.27	0.89
Other	392 (6)	35 (4)	0.67	0.47–0.96	0.03*
Comorbidity score
0	5084 (81)	623 (76)	1	Reference
1	923 (14)	127 (15)	1.12	0.91–1.38	0.26
≥2	304 (5)	70 (9)	1.88	1.43–2.47	<0.0001*
Age, y
≤65	3109 (49)	274 (33)	1	Reference
>65	3202 (51)	546 (67)	1.93	1.66–2.26	<0.0001*
Insurance
None	348 (6)	42 (5)	1	Reference
Private payer	2103 (33)	174 (21)	0.69	0.48–0.97	0.04*
Government	3757 (59)	590 (72)	1.30	0.93–1.81	0.12
Unknown	103 (2)	14 (2)	1.13	0.59–2.14	0.71
Education, %
≥29.0	1330 (21)	165 (21)	1	Reference
20.0 to 28.9	1708 (27)	210 (26)	0.99	0.80–1.23	0.94
14.0 to 19.9	1922 (30)	273 (34)	1.14	0.93–1.40	0.20
<14.0	1280 (22)	158 (21)	0.99	0.79–1.25	0.97
Facility type
Community cancer program	837 (14)	125 (15)	1	Reference
Comprehensive cancer program	2670 (44)	414 (51)	1.04	0.84–1.29	0.40
Academic/research program	2609 (42)	270 (34)	0.69	0.55–0.87	0.0015*
Facility location
Metro	5024 (82)	639 (80)	1	Reference
Urban	952 (16)	129 (16)	1.07	0.87–1.30	0.54
Rural	136 (2)	27 (4)	1.56	1.02–2.38	0.04*
Income, US $
<30,000	1301 (21)	169 (21)	1	Reference
30,000–35,000	1458 (23)	205 (25)	1.08	0.87–1.34	0.47
35,000–45,999	1664 (27)	210 (26)	0.97	0.78–1.20	0.79
>46,000	1814 (29)	220 (28)	0.93	0.75–1.16	0.53
Stage group
1	695 (11)	141 (17)	1	Reference
2	2915 (46)	401 (49)	0.68	0.55–0.84	0.0003*
3	2701 (43)	278 (34)	0.50	0.41–0.63	<0.0001*
Clinical T stage
1	321 (5)	60 (8)	1	Reference
2	616 (10)	111 (15)	0.96	0.68–1.36	0.21
3	4444 (74)	490 (64)	0.59	0.44–0.79	0.0004*
4	594 (11)	102 (13)	0.93	0.69–1.26	0.66
Distance to facility, miles
≤8.5	3085 (49)	432 (53)	1	Reference
>8.5	3226 (51)	388 (47)	0.86	0.74–0.99	0.04*
Year of diagnosis
2004–2006	1138 (18)	189 (23)	1	Reference
2007–2009	1553 (25)	237 (29)	0.92	0.75–1.13	0.42
2010–2012	1875 (30)	226 (28)	0.73	0.59–0.89	0.0024*
2013–2014	1745 (27)	168 (20)	0.58	0.46–0.72	<0.0001*

Education is quartiles of the percentage of persons with less than a high school education in the patients’ residence census tract. Income is median household income in the patients’ residence census tract.

^*P value is significant.

The median follow-up time was 31 months (range, 2–154 mo). Three- and 5-year survival rates for all of the patients were 60% and 46%. Patients treated with dose escalation had significantly worse survival compared with those treated to ≤54 Gy (Fig. 2). The median overall survival for patients treated to ≤54 Gy was 56 months compared with 33 months for higher doses (p = 0.0001). We also conducted Kaplan–Meier analysis using dose cutoffs of 49 Gy and 59 Gy, showing similar inferior outcomes in the higher dose groups. Multivariable Cox regression revealed increased age, increasing comorbidity score, urban and rural locations, dose-escalated radiation, increasing stage, and higher grade tumors to be associated with decreased survival. Multivariable Cox regression also revealed female sex, type of insurance, increased income, increasing distance to facility, and more recent year of treatment to be predictive of improved survival (Table 3). A second multivariable Cox proportional hazards model was used including the propensity score. The propensity score–adjusted multivariable analysis identified increased radiation dose and urban and rural locations as predictors for increased risk of death. Increased distance to facility and higher income correlated with improved outcomes (Table 4).

FIGURE 2.

Overall survival by dose of radiation. Median survival was 56 months vs 33 months in favor of lower doses of radiation p < 0.0001.

Table 3.

Multivariable Cox proportional hazards models for overall survival in patients receiving definitive chemoradiotherapy for rectal cancer

Significant characteristic	Hazard of death (95% CI), Cox model without propensity score	p
Age, y
≤65	Reference
>65	1.72 (1.58–1.87)	<0.0001
Distance to facility, miles
≤8.5	Reference
>8.5	0.89 (0.82–0.96)	0.002
Comorbidity score
0	Reference
1	1.28 (1.17–1.40)	<0.0001
≥ 2	1.83 (1.61–2.09)	<0.0001
Location
Metro	Reference
Urban	1.18 (1.07–1.30)	0.0011
Rural	1.40 (1.13–1.72)	0.0016
Radiation dose, Gy
≤54	Reference
>54	1.34 (1.22–1.47))	<0.0001
Insurance
None	Reference
Private	0.54 (0.46–0.63)	<0.0001
Government	0.77 (0.66–0.90)	0.0011
Sex
Men	Reference
Women	0.91 (0.86–0.98)	0.0148
Stage group
1	Reference
2	1.27 (1.14–1.42)	<0.0001
3	1.43 (1.28–1.61)	<0.0001
Income, US $
<30,000	Reference
30,000–35,000	0.95 (0.86–1.06)	0.36
35,000–45,999	0.91 (0.82–1.02)	0.12
>46,000	0.84 (0.78–0.91)	<0.0001
Grade
1 (well differentiated)	Reference
2 (moderately differentiated)	1.03 (0.91–1.18)	0.58
3 (poorly differentiated)	1.55 (1.39–1.74)	<0.0001
4 (undifferentiated)	0.96 (0.54–1.71)	0.88
Years of diagnosis
2004–2006	Reference
2007–2009	0.97 (0.89–1.07)	0.62
2010–2012	1.01 (0.92–1.10)	0.85
2013–2014	0.88 (0.77–0.94)	0.0009

Table 4.

Multivariable Cox proportional hazards models with propensity score for overall survival in patients receiving definitive chemoradiotherapy for rectal cancer

Significant characteristic	Hazard of death (95% CI), Cox model with propensity score	p
Distance to facility, miles
≤8.5	Reference
>8.5	0.85 (0.81–0.0.93)	<0.0001
Radiation, Gy
≤54	Reference
>54	1.48 (1.35–1.62)	<0.0001
Income, US $
<30,000	Reference
30,000–35,000	0.94 (0.84–1.04)	0.22
35,000–45,999	0.88 (0.78–0.98)	0.0246
>46,000	0.82 (0.76–0.89)	<0.0001
Location
Metro	Reference
Urban	1.22 (1.11–1.34)	0.0001
Rural	1.50 (1.35–1.62)	0.0001

DISCUSSION

The optimal treatment for rectal adenocarcinoma depends on many factors, including disease extent (ie, stage), symptomatology, location, and patient compliance and willingness. Regardless, surgery remains the only acknowledged curative treatment for rectal cancer and should be incorporated into the treatment algorithm whenever feasible. For patients with locally advanced disease it is well known and preferable to begin treatment with a neoadjuvant approach, including radiation and fluoropyrimidine-based chemotherapy. Several randomized trials, including the landmark trial by Sauer et al,^3,4 showed improved local control and decreased toxicity, with similar rates of overall survival when treating preoperatively. Interestingly, the complete response rate after neoadjuvant therapy in the German trial was relatively low, at 8% overall. The French FFCD 9203 trial compared preoperative radiation alone (45 Gy) with radiation with concurrent bolus 5-fluorouracil and similarly showed a pathologic complete response rate of 11.7%,¹⁴ acknowledging that 45 Gy is on the lower end of neoadjuvant doses. Likewise, the European Organization for Research and Treatment of Cancer 22921 trial showed a pathologic complete response rate of 13% with the addition of chemotherapy to preoperative radiation therapy at a dose of 45 Gy.¹⁵

Based on the above, preoperative therapy is highly preferred and remains the standard of care. A recent NCDB analysis reflects that standard and showed that the use of neoadjuvant therapy has steadily increased since 2004. At the time of publication, neoadjuvant therapy use had risen from 42% in 2004 to 55% in 2012.⁵ Patients treated with definitive chemoradiation were certainly the minority but slightly increased over that time interval from 9% to 12%. Comparing the different approaches, overall survival was significantly better in the neoadjuvant cohort compared with the definitive chemoradiation cohort, 5-year survival rates being 72% and 49%. Because those outcomes from the various treatment methods were already compared in that article, we did not seek to directly compare them in this NCDB analysis. Instead, we chose to focus on nuances in the radiation doses to see if any differences in outcome emerged.

As discussed above, the complete response rate at time of surgery after neoadjuvant chemoradiation to the 45 to 50 Gy appears to hover around the 10% to 20% range. More recently, the watch-and-wait approach has been gaining traction, which postulates that perhaps with increased radiation dose, extended chemotherapy, careful patient selection, and very close surveillance, surgery may be avoided. The Brazilians have blazed the trail for this technique using dose-escalated radiation (54 Gy), 6 cycles of 5-fluorouracil every 3 weeks followed by tumor assessment at 8 to 10 weeks.⁸ Results from their approach are quite impressive, with 50% of patients not requiring any surgical intervention and documentation of successful surgical salvage in >90% of patients, yielding an 80% rate of overall organ preservation.⁹ In addition, a group from the United Kingdom also performed a propensity score–matched cohort analysis on a subset of patients (129 total) managed at a tertiary cancer center with a watch-and-wait approach after clinical complete response. On comparison with patients treated with surgery, the 3-year nonregrowth disease-free survival rates were 88% and 78%, with no difference in overall survival.⁷

The results presented here are certainly limited by their retrospective nature and inherent selection bias present in all NCDB studies. In addition, the NCDB lacks important treatment details and outcomes as they relate to a definitive chemoradiation approach in rectal cancer, namely number of cycles of chemotherapy, type of chemotherapy, treatment-related toxicity, and, most importantly, local failure. There are some data to suggest that there is a highly significant dose-response relationship for rectal adenocarcinoma, including a study of 222 patients treated with chemotherapy and varying doses of radiation showing improved tumor regression for doses in the 50.4- to 70.0-Gy range.¹⁶ This notion is certainly supported by the data referenced above using the watch-and-wait approach, which typically uses doses of ≥54 Gy. It was for that reason that we chose that dose as a break point for comparison of outcomes. Given the detriment noted, we also analyzed groups based on doses >49 Gy and >59 Gy, with both analyses showing consistently worse overall survival with dose escalation. One should keep in mind, as alluded to above, that local control is one of the most important outcomes to track with a definitive chemoradiation approach and that data are not tracked in the NCDB. Similarly, salvage therapy is not included, which, in a watch-and-wait patient population would be surgical resection. It should also be noted that, in our series, patients who were older and had a higher comorbidity score were more likely to get dose-escalated radiation, and perhaps this more frail population had competing comorbidities leading to their ultimate demise. In addition, patients treated at an academic center were also less likely to undergo dose-escalated therapy and perhaps had access to a higher level of care (or ultimately surgical salvage), leading to a better outcome. As alluded to above, toxicity data are not captured in the NCDB, and one could also speculate that dose escalation may have led to significant toxicity in a more frail population and overall worse outcomes. Along those lines, there are strong randomized data from other disease sites (Radiation Therapy Oncology Group 0617 and Radiation Therapy Oncology Group 9405) showing worse outcomes with dose escalation in nonsmall cell lung cancer and esophageal cancer.^17,18 Thus, the concept of more dose not always being better is not new or unique.

CONCLUSION

The definitive management of rectal cancer with chemoradiation remains prevalent either because of medical necessity or perhaps by patient choice. In such cases it stands to reason that dose escalation to the primary may offer better disease control and consequently superior outcomes; however, the results presented here show that dose-escalated radiation was not associated with longer overall survival but instead a detriment. Just as caution must be used when interpreting these results, careful and appropriate patient selection (ideally on a clinical trial) should be used when attempting to provide definitive chemoradiation in lieu of a more conventional neoadjuvant approach.