Abstract
Objective:
To identify the survival benefit of different adjuvant approaches and factors influencing their efficacy after upfront resection of pancreatic ductal adenocarcinoma (PDAC).
Summary Background Data:
The optimal adjuvant approach for PDAC remains controversial.
Methods:
Patients from the National Cancer Database who underwent upfront PDAC resection from 2010 to 2014 were analyzed to determine clinical outcomes of different adjuvant treatment approaches, stratified according to pathologic characteristics. Factors associated with overall survival were identified with multivariable logistic regression and Cox proportional hazards were used to compare overall survival of different treatment approaches in the whole cohort, and propensity score matched groups.
Results:
We included 16,709 patients who underwent upfront resection of PDAC. On multivariable analysis, tumor size, grade, positive margin, nodal involvement, lymphovascular invasion (LVI), stage, lymph node ratio, not receiving chemotherapy, and/or radiation were predictors for worse survival. In the presence of at least 1 high-risk pathologic feature (nodal or margin involvement or LVI) chemotherapy with subsequent radiation provided the most significant survival benefit (median survivals: 24.8 vs 21.0 mo for adjuvant chemotherapy; HR = 0.81; 95% CI: 0.77–0.86; P < 0.001 in propensity score matching). The addition of radiation to adjuvant chemotherapy did not significantly improve overall survival in those with no highrisk pathologic features (median survivals: 54.6 vs 42.7 mo for adjuvant chemotherapy; HR=0.90; 95% CI: 0.75–1.08; P = 0.25 in propensity score matching).
Conclusions:
In the presence of any high-risk pathologic features (nodal or margin involvement or LVI), adjuvant chemotherapy followed by radiation provides a better survival advantage over chemotherapy alone after upfront resection of PDAC.
Keywords: adjuvant, pancreatic cancer, upfront resection
Pancreatic adenocarcinoma is projected to surpass colorectal cancer in cancer-related mortality and become the second most lethal cancer, within the next few years and before 2030.1 According to annual cancer statistics reported by the American Cancer Society, pancreatic cancer has the lowest overall survival rate with 5-year survival of 8%.2 While surgical resection is the only potentially curative modality, the majority of patients are diagnosed with advanced unresectable stages. Even with resection, overall survival is poor, and 5-year survival only improves to 25%.3 Moreover, a considerable number of patients, while eligible, never receive the optimal treatments, that is, surgical resection and systemic treatment.4 Systemic treatment is an essential component of a comprehensive treatment plan for patients with pancreatic adenocarcinoma.5 This is primarily attributable to the aggressive nature of pancreatic cancer with early systemic dissemination and high-risk for local failure and distant recurrence.6
Multimodality treatment of pancreatic cancer has been evolving in the past few years, yet there are significant questions that remain to be answered. While standard treatment for pancreatic cancer is surgical resection (when operable) followed by adjuvant chemotherapy,7–9 standard of care has been challenged with a variety of alternative approaches, including variations in sequencing of resection and systemic treatment, as well as the role of radiation in the neoadjuvant or adjuvant setting. Even in the patients who undergo upfront surgery, the role of radiation in the adjuvant setting has not been well established, and guidelines do not provide a strict recommendation with this regard. This is attributed to conflicting data in the literature, mainly due to poorly designed and performed earlier clinical trials.10
In this study, we investigated the National Cancer Database (NCDB) to understand the practice patterns and outcomes of different adjuvant approaches better. We stratified patients based on various pathologic features and studied different adjuvant approaches and compared the outcomes to identify outcome-based optimal treatment approach according to clinicopathologic features.
METHODS
Study Population
Records were obtained from NCDB of the American College of Surgeons Commission on Cancer. NCDB is a large retrospective database from 1500 Commission-accredited facilities that include approximately 74% of patients with pancreatic cancer.11 To focus on more recent practice patterns, we excluded patients who were treated before 2010 and selected all the patients with a diagnosis of pancreatic adenocarcinoma from 2010 to 2014 who underwent upfront resection of pancreatic adenocarcinoma.
Study Variables
We included the histology codes 8140, 8148, 8453, 8503, 8470, 8500, 8560, 8576, 8510, 8480, 8490, 8020, 8035, 8550, 8551, 8453, 8503, 8470, 8971, 8441, 8452, 8246, 8041, 8013, 8552, and 8154. We studied baseline clinicopathologic characteristics including age, gender, race, comorbidities (Charlson/Deyo score), tumor size, tumor grade of differentiation, AJCC 7th edition pathologic stage and separately analyzed for T-category, N-category, margin status, number of examined lymph nodes, number of positive lymph nodes, and lymphovascular invasion (LVI). Lymph node ratio was calculated as the number of positive lymph nodes divided by the number of examined lymph nodes. The interval between surgery and adjuvant treatment was calculated by subtracting the number of days between diagnosis and initiation of adjuvant treatment (chemotherapy or radiation therapy) from the number of days between diagnosis and the date of surgery.
All the patients with stage 4 were excluded from the study. All the included patients had pancreatic adenocarcinoma (PDAC) histology, and patients with pancreatic neuroendocrine tumors and metastatic pancreatic tumors were excluded. Adjuvant chemotherapy was considered an interval of less than 180 days after surgery. Adjuvant radiation was considered a range of less than 180 days after surgery or adjuvant chemotherapy.
Based on the treatment modality and the sequence of treatment, adjuvant treatment approaches were categorized into 5 groups: no treatment, chemotherapy only (C), radiation alone (RT), chemotherapy followed by radiation (C-RT), radiation followed by chemotherapy (RT-C).
Overall survival was considered the primary endpoint. Outcomes (overall survival) were adjusted for pathological and clinical characteristics including gender, age, facility type, comorbidities, type of adjuvant treatment, tumor size, grade, pathologic stage, lymph node ratio, lymphovascular invasion, nodal involvement, and margin status.
Statistical Analysis
Descriptive analysis of categorical variables was calculated as counts with proportions and for continuous variables as means and standard deviations. For study group comparisons, we used independent t tests for continuous data with 2 group comparisons and analysis of variance with more than 2 groups and chi-squared tests for categorical data. Survival estimates were calculated using the Kaplan–Meier method. Group differences in unadjusted survival models were calculated using log-rank tests; to estimate the hazard for clinicopathologic features, Cox proportional hazards regression modeling was used in univariate and multivariate regression analysis. Propensity score (PS) matching was performed using logistic regression to model the probability of receiving chemotherapy or chemotherapy followed by radiation as a function of the potential confounding variables. A balanced distribution of PS-matched patients in the 2 groups was validated using independent t tests for continuous variables and chi-squared tests for categorical variables. A P value of <0.05 was considered statistically significant. SAS (version 9.4, Cary, NC) was used for statistical analyses.
RESULTS
A total of 16,709 patients who underwent upfront resection of PDAC were selected. Among them, 10,825 (64.8%) patients received some type of adjuvant treatment, while 5884 (35.2%) did not receive any adjuvant treatment. Tables 1 and 2 illustrate the demographic (Table 1) and clinicopathologic features (Table 2) of the cohort. Overall, 10,680 patients (63.9%) received postoperative chemotherapy and 4924 patients (29.5%) received postoperative radiation. The majority of patients had a moderately differentiated tumor (51.4%), were stage 2 (84.7%), and had nodal involvement (67.4%) and lymphovascular invasion (51.0%). Positive margin was reported in 3845 (23.4%) patients. Overall median survival was 22.1 months and overall median follow-up was 42.9 months. The most common adjuvant treatment approach was adjuvant chemotherapy in 5901 (35.4%) of the patients, followed by C-RT in 4488 (26.9%), RT-C in 291 (1.8%), and RT in 145 patients (0.9%).
TABLE 1.
Demographic Features of Patients Who Received Various Adjuvant Treatments After Resection of Pancreatic Cancer
| Variable | Total | None | RT | Chemo | RT-chemo | Chemo-RT | P Value |
|---|---|---|---|---|---|---|---|
| Age | 65.7±10.8 | 68.0±13.3 | 67.3±12.0 | 65.3±10.3 | 64.4±10.0 | 63.2±9.5 | <0.001 |
| Sex | |||||||
| Male | 8529 (51.0%) | 2881 (49.0%) | 86 (59.3%) | 2967 (50.3%) | 147 (50.5%) | 2448 (54.5%) | <0.001 |
| Female | 8180 (49.0%) | 3003 (51.0%) | 59 (40.7%) | 2934 (49.7%) | 144 (49.5%) | 2040 (45.5%) | |
| Race/ethnicity | |||||||
| White | 14,124 (84.5%) | 4929 (83.7%) | 116 (80.0%) | 5019 (85.1%) | 237 (%)81.4 | 3823 (85.2%) | <0.001 |
| Black | 1796 (10.7%) | 655 (11.1%) | 23 (15.8%) | 600 (10.2%) | 38 (13.1%) | 480 (10.7%) | |
| Academic | |||||||
| No | 7623 (46.4%) | 2517 (44.05%) | 86 (60.56%) | 2472 (42.33%) | 174 (60%) | 2374 (53.43%) | <0.001 |
| Yes | 8806 (53.6%) | 3197 (55.95%) | 56 (39.44%) | 3368 (57.67%) | 116 (40%) | 2069 (46.57%) | |
| Location | |||||||
| Northeast | 3423 (20.8%) | 1121 (18.1%) | 27 (18.1%) | 1493 (23.5%) | 48 (16%) | 938 (20.6%) | <0.001 |
| Southeast | 3730 (22.7%) | 1423 (22.9%) | 38 (25.5%) | 1302 (20.5%) | 86 (28.7%) | 1118 ((24.6%) | |
| East central | 4155 (25.3%) | 1527 (24.7%) | 52 (34.9%) | 1598 (25.2%) | 90 (30%) | 1201 (26.4%) | |
| West central | 2863 (17.4%) | 1210 (19.5%) | 18 (12.1%) | 1039 (16.7%) | 48 (16%) | 724 (15.9%) | |
| West coast | 2258 (13.7%) | 909 (14.7%) | 14 (9.4%) | 917 (14.4%) | 28 (9.3%) | 566 (12.5%) | |
| Insurance | |||||||
| Not Insured | 590 (3.4%) | 215 (3.7%) | 10 (7.04%) | 174 (2.98%) | 12 (4.14%) | 133 (2.99%) | <0.001 |
| Private | 6597 (37.9%) | 1634 (28.15%) | 36 (25.35%) | 2264 (38.8%) | 108 (37.24%) | 2078 (46.77%) | |
| Medicaid | 981 (5.6%) | 331 (5.7%) | 10 (7.04%) | 315 (5.4%) | 25 (8.62%) | 238 (5.36%) | |
| Medicare | 9236 (53.1%) | 3549 (61.15%) | 83 (58.45%) | 3025 (51.84%) | 143 (49.31%) | 1914 (43.08%) | |
| Income | |||||||
| <63K | 10,959 (66.0%) | 4069 (69.59%) | 106 (73.1%) | 3678 (62.65%) | 220 (76.12%) | 2886 (64.67%) | <0.001 |
| 63Kþ | 5656 (34.0%) | 1778 (30.41%) | 39 (26.9%) | 2193 (37.35%) | 69 (23.88%) | 1577 (35.33%) | |
| Education | |||||||
| > 21% | 2726 (16.4%) | 1156 (19.76%) | 32 (22.07%) | 832 (14.16%) | 58 (20.07%) | 648 (14.51%) | <0.001 |
| (No HSD) | |||||||
| < 21% | 13,896 (83.6%) | 4693 (80.24%) | 113 (77.93%) | 5042 (85.84%) | 231 (79.93%) | 3817 (85.49%) | |
| Comorbidities | |||||||
| No | 10,849 (64.9%) | 3693 (62.76%) | 86 (59.31%) | 3798 (64.36%) | 187 (64.26%) | 3085 (68.74%) | <0.001 |
| Yes | 5860 (35.1%) | 2344 (36.8%) | 62 (40.79%) | 2282 (35.51%) | 108 (35.76%) | 1432 (31.18%) | |
All the treatments are in the adjuvant setting.
Chemo indicates chemotherapy; Chemo-RT, chemotherapy followed by radiation; HSD, high-school degree; RT, radiation; RT-Chemo, radiation followed by chemotherapy.
TABLE 2.
Clinicopathologic Features of Patients Who Received Various Adjuvant Treatments After Resection of Pancreatic Cancer
| Variable | Total | None | RT | Chemo | RT-chemo | Chemo-RT | P Value |
|---|---|---|---|---|---|---|---|
| Median survival (m) | 22.1 | 18.9 | 10.0 | 30.3 | 22.2 | 36.4 | <0.001 |
| Median Follow-up (m) | 42.9 | 41.6 | 51.5 | 40.8 | 43.0 | 46.4 | <0.001 |
| Grade | |||||||
| 1 | 1677 (10.3%) | 680 (12.13%) | 18 (13.43%) | 520 (8.75%) | 35 (12.59%) | 424 (9.87%) | <0.001 |
| 2 | 8359 (51.4%) | 2793 (49.81%) | 66 (49.25%) | 3085 (51.88%) | 131 (47.12%) | 2284 (53.18%) | |
| 3 | 5995 (36.9%) | 2044 (36.45%) | 48 (35.82%) | 2254 (37.91%) | 108 (38.85%) | 1541 (35.88%) | |
| 4 | 229 (1.4%) | 90 (1.61%) | 2 (1.49%) | 87 (1.46%) | 4 (1.44%) | 46 (1.07%) | |
| Stage | |||||||
| 1 | 2011 (12.0%) | 1059 (18%) | 14 (9.66%) | 623 (10.56%) | 16 (5.5%) | 299 (6.66%) | <0.001 |
| 2 | 14145 (84.7%) | 4634 (78.76%) | 122 (84.14%) | 5108 (86.56%) | 263 (90.38%) | 4018 (89.53%) | |
| 3 | 553 (3.3%) | 191 (3.25%) | 9 (6.21%) | 170 (2.88%) | 12 (4.12%) | 171 (3.81%) | |
| Nodes | |||||||
| Negative | 5301 (32.6%) | 2268 (40.11%) | 48 (34.29%) | 1777 (30.75%) | 72 (25.71%) | 1136 (25.73%) | <0.001 |
| Positive | 10,967 (67.4%) | 3386 (59.89%) | 92 (65.71%) | 4002 (69.25%) | 208 (74.29%) | 3279 (74.27%) | |
| Lymph node ratio | 0.170.20 | 0.160.21 | 0.160.19 | 0.170.20 | 0.200.21 | 0.180.21 | <0.001 |
| LVI | |||||||
| No | 7117 (49.0%) | 2750 (53.6%) | 64 (56.14%) | 2362 (45.91%) | 123 (47.67%) | 1818 (46.98%) | <0.001 |
| Yes | 7401 (51.0%) | 2381 (46.4%) | 50 (43.86%) | 2783 (54.09%) | 135 (52.33%) | 2052 (53.02%) | |
| Tumor size | 35.5±32.7 | 35.8±34.6 | 35.3±15.8 | 35.0±29.2 | 37.0±22.8 | 35.7±35.5 | 0.59 |
| T | |||||||
| 1 | 975 (6.0%) | 495 (8.58%) | 8 (5.63%) | 327 (5.61%) | 11 (3.86%) | 134 (3.03%) | <0.001 |
| 2 | 2003 (12.3%) | 825 (14.3%) | 14 (9.86%) | 674 (11.56%) | 20 (7.02%) | 470 (10.63%) | |
| 3 | 12,918 (79.3%) | 4223 (73.18%) | 110 (77.46%) | 4671 (80.12%) | 248 (87.02%) | 3666 (82.88%) | |
| 4 | 385 (2.7%) | 135 (2.34%) | 7 (4.93%) | 108 (1.85%) | 6 (2.11%) | 129 (2.92%) | |
| Margin | |||||||
| No | 12,568 (76.6%) | 4496 (77.83%) | 88 (63.31%) | 4707 (81.25%) | 177 (62.11%) | 3100 (70.15%) | <0.001 |
| Yes | 3845 (23.4%) | 1281 (22.17%) | 51 (36.69%) | 1086 (18.75%) | 108 (37.89%) | 1319 (29.85%) | |
All the treatments are in the adjuvant setting.
Chemo indicates chemotherapy; Chemo-RT, chemotherapy followed by radiation; LVI, lymphovascular invasion; XRT, radiation; RT-Chemo, radiation followed by chemotherapy
Factors Associated With Overall Survival
Table 3 illustrates the univariate and multivariate analyses of factors associated with overall survival. Among the pathologic factors, higher grade of differentiation, nodal involvement, margin involvement, LVI, lymph node ratio, tumor size, and higher stages were independently associated with worse overall survival. Administration of chemotherapy (OR = 1.53; 95% CI 1.46–1.61, P < 0.001) and to a lesser extent, radiation (OR = 1.18; 95% CI: 1.12–1.24, P < 0.001) were independently associated with improved overall survival. When overall chemotherapy and radiation therapy was replaced with specific adjuvant approaches in the multivariate model, adjuvant C (OR = 0.65; 95% CI 0.62–0.69), C-RT (OR = 0.54; 95% CI: 0.51–0.57), and RT-C (OR = 0.71; 95% CI: 0.63– 0.81) were associated with improved overall survival compared with no adjuvant treatment. Adjuvant RT did not improve overall survival compared with no treatment (HR = 0.87; 95% CI: 0.70–1.09). Compared with adjuvant C (standard of care), C-RT (HR = 0.83; 95% CI: 0.79–0.88) was associated with improved survival while RT was associated with worse survival (HR = 1.34; 95% CI: 1.07–1.67) and overall survival was not significantly different from RT-C (HR = 1.09; 95% CI: 0.93–1.27). Comparison of the impact of various adjuvant treatments on overall survival of patients who undergo upfront resection of pancreatic cancer in the whole cohort is demonstrated in Table 4A and Kaplan–Meier survival curve comparing various adjuvant approaches in the entire cohort is illustrated in Figure 1A. Best overall survival is observed with chemotherapy followed by radiation and administration of radiation before chemotherapy results in worse survival.
TABLE 3.
Factors Associated With Overall Survival in Patients Who Underwent Upfront Pancreatic Cancer Resection
| Variable | Univariate |
Multivariate |
||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P Value | HR | 95% CI | P Value | |
| Age | ||||||
| (per 1 yr) | 1.02 | 1.01–1.02 | <0.001 | 1.01 | 1.01–1.01 | <0.001 |
| Sex | ||||||
| Female | 0.93 | 0.90–0.97 | <0.001 | 1.01 | 0.96–1.01 | 0.660 |
| Academic center | ||||||
| No | 1.19 | 1.14–1.23 | <0.001 | 1.15 | 1.10–1.20 | <0.001 |
| Comorbidities | ||||||
| Yes | 1.12 | 1.08–1.16 | <0.001 | 1.11 | 1.06–1.16 | <0.001 |
| Grade (vs 1/2) | ||||||
| 3/4 | 1.43 | 1.37–1.49 | <0.001 | 1.36 | 1.30–1.42 | <0.001 |
| Stage | ||||||
| 2 | 2.37 | 2.21–2.55 | <0.001 | 1.46 | 1.29–1.65 | <0.001 |
| (vs 1) | ||||||
| 3 | 3.73 | 3.00–4.18 | <0.001 | 1.71 | 1.28–2.27 | <0.001 |
| Nodal involvement | ||||||
| Yes | 1.86 | 1.79–1.95 | <0.001 | 1.28 | 1.20–1.37 | <0.001 |
| Margin | ||||||
| Involved | 1.68 | 1.62–1.76 | <0.001 | 1.48 | 1.41–1.56 | <0.001 |
| LVI | ||||||
| Present | 1.54 | 1.48–1.61 | <0.001 | 1.16 | 1.10–1.21 | <0.001 |
| Lymph node ratio | ||||||
| (Unit 1) | 1.14 | 1.13–1.15 | <0.001 | 1.86 | 1.64–2.11 | <0.001 |
| Tumor size | ||||||
| (Unit 10 mm) | 1.01 | 1.01–1.02 | <0.001 | 1.01 | 1.01–1.02 | <0.001 |
| Chemotherapy | ||||||
| No | 1.37 | 1.31–1.42 | <0.001 | 1.53 | 1.46–1.61 | <0.001 |
| Radiation | ||||||
| No | 1.21 | 1.16–1.26 | <0.001 | 1.18 | 1.12–1.24 | <0.001 |
LVI indicates lymphovascular invasion
TABLE 4.
Comparison of the Impact of Various Adjuvant Treatments on Overall Survival of Patients Who Undergo Resection of Pancreatic Cancer Tumor
| A | |||||
|---|---|---|---|---|---|
|
| |||||
| Chemo | Chemo-RT | RT-chemo | RT | None | |
| Chemo | NA | 1.09 (1.04–1.15) | 0.79 (0.69–0.91) | 0.78 (0.64–0.93) | 0.76 (0.72–0.79) |
| Chemo-RT | 0.92 (0.87–0.96) | NA | 0.73 (0.63–0.83) | 0.71 (0.58–0.86) | 0.69 (0.66–0.73) |
| RT-Chemo | 1.27 (1.1–1.45) | 1.37 (1.2–1.59) | NA | 0.98 (0.78–1.23) | 0.95 (0.83–1.1) |
| RT | 1.29 (1.07–1.56) | 1.41 (1.16–1.71) | 1.02 (0.81–1.29) | NA | 0.97 (0.81–1.18) |
| None | 1.32 (1.27–1.39) | 1.45 (1.37–1.52) | 1.05 (0.91–1.2) | 1.03 (0.85–1.23) | NA |
| B | |||||
|
| |||||
| Chemo | Chemo-RT | RT-chemo | RT | None | |
|
| |||||
| Chemo | NA | 1.18 (1.12–1.24) | 0.83 (0.72–0.97) | 0.77 (0.63–0.94) | 0.64 (0.61–0.67) |
| Chemo-RT | 0.85 (0.81–0.89) | NA | 0.71 (0.61–0.82) | 0.65 (0.53–0.81) | 0.54 (0.51–0.57) |
| RT-Chemo | 1.2 (1.03–1.39) | 1.41 (1.22–1.64) | NA | 0.93 (0.72–1.19) | 0.76 (0.66–0.88) |
| RT | 1.3 (1.06–1.59) | 1.53 (1.24–1.88) | 1.08 (0.84–1.39) | NA | 0.82 (0.67–1.01) |
| None | 1.56 (1.49–1.64) | 1.85 (1.75–1.96) | 1.31 (1.13–1.52) | 1.22 (0.99–1.49) | NA |
| C | |||||
|
| |||||
| Chemo | Chemo-RT | RT-chemo | XRT | None | |
|
| |||||
| Chemo | NA | 1.01 (0.86–1.18) | 0.75 (0.48–1.17) | 0.64 (0.35–1.16) | 0.87 (0.76–0.98) |
| Chemo-RT | 0.99 (0.85–1.16) | NA | 0.74 (0.47–1.17) | 0.63 (0.35–1.16) | 0.86 (0.74–0.99) |
| RT-Chemo | 1.34 (0.85–2.09) | 1.35 (0.86–2.13) | NA | 0.85 (0.41–1.78) | 1.16 (0.74–1.81) |
| RT | 1.56 (0.86–2.84) | 1.58 (0.86–2.88) | 1.17 (0.56–2.44) | NA | 1.35 (0.74–2.45) |
| None | 1.16 (1.02–1.31) | 1.17 (1.01–1.35) | 0.87 (0.55–1.35) | 0.74 (0.41–1.34) | NA |
A: Whole cohort; B: Patients with margin and/or node-positive and/or lymphovascular invasion; C: Patients with negative margin and node and no lymphovascular invasion.
FIGURE 1.
Kaplan–Meier survival curves comparing various adjuvant approaches after upfront surgery. (A) In the whole cohort; (B) in patients with at least 1 high-risk pathologic feature (positive node and/or margin involvement and/or lymphovascular invasion); (C) in patients with no high-risk pathologic features (no nodal and margin involvement and no lymphovascular invasion).
Impact of Adjuvant Approach on Overall Survival Stratified by Pathologic Features
To investigate whether pathologic features have any differential impact on relative overall survival of different adjuvant approaches, we stratified patients based on important pathologic factors that are independently associated with overall survival. The distribution of high-risk pathologic features is provided in the supplemental Table 1, http://links.lww.com/SLA/B915. In patients with nodal involvement (supplemental Figure 1–A, http://links.lww.com/SLA/B918), positive margin (supplemental Figure 1–B, http://links.lww.com/SLA/B918), or the presence of LVI (supplemental Figure 1–C, http://links.lww.com/SLA/B918), best overall survival was observed with C-RT, while in patients with negative node (supplemental Figure 1–D, http://links.lww.com/SLA/B918) and absence of LVI (supplemental Figure 1–F, http://links.lww.com/SLA/B918) there was no difference between overall survival with adjuvant C-RTand adjuvant C. In patients without margin involvement (supplemental Figure 1–E, http://links.lww.com/SLA/B918), adjuvant C-RT was still superior to adjuvant C. To further investigate whether each pathologic feature would be sufficient to stratify patients’ response to adjuvant treatment, in the subset of node-negative patients we selected margin positive group (node −/margin+) and observed a survival benefit for C-RT compared with C. (HR = 0.81; 95% CI 0.72–0.90; P < 0.001)(Supplemental Figure 2A, http://links.lww.com/SLA/B919). Similarly, in the subset of node-negative and margin negative patients, we selected those with present LVI (LVI+/node−/margin−) and showed a similar survival advantage for C-RT compared with C (HR = 0.88; 95% CI 0.78–0.99; P = 0.03) (Supplemental Figure 2–B, http://links.lww.com/SLA/B919). Another important observation in this group is the significant adverse impact of LVI on survival (Median OS was 27.9 months (LVI+/node−/margin−) versus 48.1 months (LVI−/node/−margin−), HR = 1.61; 95% CI 1.46–1.78; P < 0.001)
Based on these findings, patients were classified into 2 groups: the group with negative high-risk pathologic features (node/margin/ LVI), and those positive for at least 1 high-risk pathologic feature (positive node/margin/LVI). Survival analysis of different adjuvant approaches was performed in each group separately and compared against each other (Table 4B and C). In patients with at least 1 positive pathologic feature, adjuvant C-RT was superior to the other adjuvant approaches. Adjuvant C was superior to all other approaches except for C-RT in patients with high-risk pathologic features (HR = 1.18; 95% CI: 1.12–1.24) (Table 4B and Fig. 1B). In patients with no high-risk pathologic features, all adjuvant approaches are superior to no adjuvant treatment, while there was no survival advantage for any adjuvant approach compared with the other treatments (Table 4C and Fig. 1C). Absence of high-risk features conferred a remarkably higher overall survival, yet there was no difference between adjuvant C and adjuvant C-RT (median OS 49.8 vs 53.6 mo; HR = 1.01; 95% CI: 0.86–1.18).
Propensity Score Matched Analysis of the Impact of Different Adjuvant Treatments on Overall Survival
To validate the above findings in a well-balanced population and minimize the confounding factors, PS matching was performed. The cohort was first divided into 2 groups: 1) Presence of at least 1 high-risk pathologic feature (positive node/margin/LVI). 2) No highrisk pathologic features (negative node/margin/LVI). PS matching with 1:1 ratio was performed for the variables listed in Table 5 and supplemental Table 2, http://links.lww.com/SLA/B916. In the first group with positive pathologic features, from 4582 who received C and 3795 who received C-RT, 3418 patients were matched in each group. A balanced distribution of demographic and clinicopathologic features is illustrated in supplemental Table 2, http://links.lww.com/SLA/B916 and Table 5. The median survivals were 21.0 months for adjuvant C and 24.8 months adjuvant C-RT (HR = 1.23; 95% CI: 1.16–1.30; P < 0.001). Survival curves are demonstrated in Figure 2A.
TABLE 5.
Clinicopathologic Features of Patients Who Received Adjuvant Chemotherapy or Chemotherapy Followed by Radiation After Resection of Pancreatic Cancer
| Categorical Variable | Chemo | Chemo-RT | P Value |
|---|---|---|---|
| Median survival (mo) | 21.0 | 24.8 | <0.0001 |
| Grade | |||
| 1 | 266 (8.3%) | 288 (8.9%) | 0.622 |
| 2 | 1692 (52.5%) | 1668 (51.8%) | |
| 3 | 1227 (38.1%) | 1235 (38.3%) | |
| 4 | 38 (1.2%) | 31 (1.0%) | |
| Stage | |||
| 1 | 79 (2.3%) | 84 (2.5%) | 0.897 |
| 2 | 3224 (94.3%) | 3223 (94.3%) | |
| 3 | 115 (3.4%) | 111 (3.2%) | |
| T | |||
| 1 | 79 (2.3%) | 79 (2.3%) | 0.999 |
| 2 | 296 (8.8%) | 297 (8.8%) | |
| 3 | 2912 (86.4%) | 2910 (86.4%) | |
| 4 | 83 (2.5%) | 84 (2.5%) | |
| Comorbidities | |||
| No | 2294 (67.1%) | 2328 (68.1%) | 0.394 |
| Yes | 1124 (32.7%) | 1090 (31.9%) | |
| Tumor size (mm) | 35.8±25.9 | 36.0±35.0 | 0.814 |
| # of examined nodes | 18.0±9.6 | 18.0±9.6 | 0.847 |
| # of positive nodes | 3.50±3.42 | 3.41±3.40 | 0.288 |
| LNR | 0.22±0.21 | 0.21±0.20 | 0.366 |
The cohort of propensity score matching of patients with at least 1 high-risk pathologic feature (margin involvement and/or nodal involvement and/or lymphovascular invasion).
All the treatments are in the adjuvant setting. Categorical variables presented in number (%) and continuous variables presented as mean standard deviation.
Chemo indicates chemotherapy; Chemo-RT, chemotherapy followed by radiation; LNR, Lymph node ratio.
FIGURE 2.
Kaplan–Meier survival curves comparing chemotherapy versus chemotherapy followed by radiation after propensity score matching: (A) in propensity score matched patients with at least 1 high-risk pathologic feature (positive node and/ or margin involvement and/or lymphovascular invasion); (B) in patients with no high-risk pathologic features (no nodal and margin involvement and no lymphovascular invasion).
In the second group with no high-risk pathologic features, from 1060 who received adjuvant C and 558 who received adjuvant C-RT, 541 patients were matched in each group. A balanced distribution of demographic and clinicopathologic features is illustrated in supplemental Table 3, http://links.lww.com/SLA/B917 and Table 6. The median survivals were 42.7 months for adjuvant C and 54.6 months for C-RT patients (for C-RT: HR = 0.90; 95% CI: 0.75–1.08; P = 0.25). Survival curves are demonstrated in Figure 2B.
TABLE 6.
Clinicopathologic Features of Patients Who Received Adjuvant Chemotherapy or Chemotherapy Followed by Radiation After Upfront Resection of Pancreatic Cancer
| Categorical Variable | Chemo | Chemo-RT | P Value |
|---|---|---|---|
| Median survival (mo) | 42.7 | 54.6 | 0.247 |
| Grade | |||
| 1 | 70 (13.6%) | 72 (14.1%) | 0.418 |
| 2 | 298 (58.0%) | 296 (57.8%) | |
| 3 | 142 (27.6%) | 134 (26.2%) | |
| 4 | 4 (0.8%) | 10 (1.9%) | |
| Stage | |||
| 1 | 173 (32.0%) | 164 (30.3%) | 0.840 |
| 2 | 367 (67.8%) | 376 (69.5%) | |
| 3 | 1 (0.2%) | 1 (0.2%) | |
| T | |||
| 1 | 52 (9.7%) | 44 (8.3%) | 0.418 |
| 2 | 118 (22.1%) | 116 (21.8%) | |
| 3 | 363 (68.0%) | 371 (69.7%) | |
| 4 | 1 (0.2%) | 1 (0.2%) | |
| Comorbidities | |||
| No | 356 (65.8%) | 361 (66.7%) | 0.748 |
| Yes | 185 (34.2%) | 180 (33.3%) | |
| Tumor size (mm) | 36.1±61.8 | 33.9±45.2 | 0.517 |
| # of examined nodes | 14.9±9.2 | 14.9±8.9 | 0.976 |
Cohort of propensity score matching of patients with no high-risk features (negative margin, no nodal involvement, and no lymphovascular invasion).
All the treatments are in the adjuvant setting. Categorical variables presented in number (%) and continuous variables presented as meanstandard deviation.
Chemo indicates chemotherapy; Chemo-RT, chemotherapy followed by radiation.
DISCUSSION
The optimal adjuvant treatment approach after upfront resection of PDAC is still an area of debate and controversy. Randomized trials (CONKO-001, ESPAC-1) have shown a survival advantage for adjuvant chemotherapy over observation, after upfront surgery for all the patients.12,13 The data, however, is conflicting regarding the role of radiation in the adjuvant setting and there is a lack of level 1 unequivocal evidence. In addition to the shortcomings and criticisms on the design of many of these studies, another explanation could be that the survival advantage is selective in a subset of patients as opposed to a global benefit. In this study, we have attempted to address the current controversies in a large patient population. We included patients who are treated after 2010 to better represent practice patterns that are similar to present-day approaches. We show that both chemotherapy and radiation therapy are associated with improved overall survival in the adjuvant setting after adjusting for clinicopathologic characteristics. While both modalities improve survival compared with observation, the survival advantage of chemotherapy followed by radiation was superior to chemotherapy alone. Other important observations were lack of survival benefit for radiation alone and the significance of the sequence of treatment. Administration of radiation before chemotherapy was associated with worse overall survival, likely due to delay in systemic therapy. To investigate whether high-risk pathologic features have any impact on patients’ benefit from adjuvant C-RT, we stratified them based on node status, margin status, and lymphovascular invasion and demonstrated that in the presence of any of these high-risk features, adjuvant C-RT has the best survival outcomes, a finding that was validated with the propensity score matching. This benefit is likely a result of better local control in the setting of these high-risk features for local recurrence. In the patients who did not have any of these high-risk pathologic features, there was no statistically significant improvement in overall survival with the addition of adjuvant radiation. Nonetheless, the median survival was higher (54.7 vs 41.1 mo) with adjuvant C-RT and survival curve shows a delayed separation and a long-term survival advantage for adjuvant C-RT. This could represent a potential benefit from adjuvant C-RT in a subset of patients and further investigation (perhaps at the molecular and personalized level) is warranted to identify patients with negative high-risk pathologic features who would potentially benefit from the addition of adjuvant radiation. Another interesting observation was a remarkable survival advantage in patients with no high-risk features. The absence of any high-risk feature was associated with an increase of more than double in the survival of patients who received adjuvant C or C-RT. While lymph node involvement and margin status are well-known risk factures for poor outcomes, the prognostic effect of LVI is less appreciated in practice and should have a greater impact in clinical decision making.
While this study was not designed to investigate the role of radiation in the neoadjuvant treatment, the survival benefit of adding radiation to chemotherapy suggests a possible role for the addition of radiation as part of the neoadjuvant approach. This hypothesis is under investigation in several ongoing studies and the results are highly anticipated.
The controversy in survival benefit of adjuvant radiation stems from negative historic trials with several flaws in their designs that raised criticisms and later contested with positive, robust retrospective data. The first trial of these series was the Gastrointestinal Tumor Study Group 9173 trial in which a total of 43 patients with who underwent resection of PDAC with negative margin were randomized to receive chemoradiation versus observation and the results showed improved survival (11 vs 20 mo, P = 0.03) with chemoradiation.14 This trial was terminated early due to poor accrual but another 30 patients were enrolled in a nonrandomized fashion into the chemoradiation arm and the results were similar.15 Subsequent clinical trials, however, did not replicate the same findings. EORTC-40891 (European Organization for Research and Treatment of Cancer) randomized 218 patients to receive adjuvant chemoradiation or undergo observation.16 While overall survival was not different between the 2 groups, the design and implementation of the trial had several flaws. The study was statistically underpowered, half of the patients had a favorable periampullary tumor and there was a lack of mandatory radiation quality assurance. Twenty percent of the patients in the treatment arm did not receive the treatment and 50% did not receive the chemotherapy according to the protocols. Nonetheless, a statistical reanalysis of this trial reported better survival in the chemoradiation arm in a 1-sided analysis.17 Another landmark trial was ESPAC-1 (European Study for Pancreatic Cancer), which was the largest prospective investigation of adjuvant therapy for pancreatic cancer. A total of 541 patients were allocated into 3 randomizations: A 22 factorial design between observation, chemotherapy, chemoradiation, and chemoradiation followed by chemotherapy, an observation versus chemotherapy randomization, and observation versus chemoradiation. An initial report showed chemotherapy was superior to observation while chemoradiation was not.18 The final results with factorial randomization demonstrated chemotherapy was superior, followed by the combination arm followed by chemoradiation.13 The conclusions drawn from ESPAC-1 were also widely criticized, mainly focused on delayed administration of chemotherapy in the combination arm. Relative survival advantage observed in our study is in fact similar to ESPAC-1 with the exception of including another combination arm in which adjuvant chemotherapy preceded radiation.
Despite all these shortcomings, the results of EORTC-40891 and ESPAC-1 resulted in the abandonment of radiation in adjuvant approaches for PDAC with upfront surgery. The benefit of adding radiation was, however later reinstated in some retrospective studies. A bi-institutional study of 1092 resected PDAC patients from Mayo Clinic and John’s Hopkins found a survival advantage for chemoradiation compared with observation (21.9 vs 14.3 mo; P < 0.001). In this study, positive margin was reported in 33% and nodal involvement was observed in 68% of patients and compared with older trials, modern radiation doses and fractions were utilized.19 Another NCDB study included patients with PDAC resection from 1998 to 2009 who received chemotherapy or chemotherapy and radiation in the adjuvant setting and showed that the addition of radiation was associated with improved overall survival.20 Compared with our study, this report represents a different patient population with no overlap, from an era in which older treatment approaches were utilized. The sequence of treatment was not studied and the combination arm included both approaches, and while subgroup analysis was performed, patient stratification was not considered to identify subgroups that may not benefit from combination therapy. Moreover, LVI as an important independent survival predictor was not studied in this analysis, or in any of the previously mentioned reports. The impact of adjuvant radiation was investigated in the Surveillance, Epidemiology, and End Results database as well and the survival benefit of radiation in nodal involvement was demonstrated. Although propensity score matching was not performed and other important pathologic features including margin status and LVI, were not studied.
This study has several limitations, mainly inherent to the retrospective nature of the database. To minimize selection bias, propensity score matching was performed and validated the same findings. A significant shortcoming of NCDB is the lack of information about specific chemotherapy regimens. The data regarding locoregional failure and progression-free survival is a valuable endpoint that is not provided in the NCDB. Although we hypothesize that the main benefit of the addition of radiation is locoregional control, the data lack appropriate endpoints to study and validate this hypothesis. Another limitation is including a broad spectrum of pancreatic adenocarcinomas with different tumor biologies. A prominent example is tumors rising from intraductal papillary mucinous neoplasms that have different biology with a lower frequency of high-risk pathologic features (less lymph node involvement and lymphovascular invasion) and better outcomes. The database does not capture the subtypes that have more favorable features (eg, colloid subtypes) and outcomes are not stratified according to various subtypes with different tumor biologies.
In conclusion, to our knowledge, this is the most comprehensive retrospective study of adjuvant treatment after upfront surgery that reflects therapeutic approaches that have been implemented nationwide in recent years. We demonstrate the independent impact of high-risk pathologic features, including nodal status, margin status and LVI on patients’ survival benefit from different adjuvant approaches. We demonstrated, in the presence of at least one high-risk pathologic feature, chemotherapy followed by radiation provides the most significant survival benefit compared with observation or other conventional approaches, including chemotherapy alone, which is the current standard of care. Although there could be some survival benefit in a subset of patients with negative high-risk pathologic features, our data do not support an overall survival advantage for this group of patients. While results of robust randomized clinical trials are awaited, this study, in addition to previous reports, provides further evidence for the efficacy of radiation as a component of a multimodality postoperative approach to improve outcomes in patients with pancreatic ductal adenocarcinoma who undergo upfront surgery. The significant role of high-risk pathologic features also provides essential information that could be instrumental in designing future clinical trials that are sufficiently powered to investigate the optimal treatment approach in patients with different pathologic characteristics.
Supplementary Material
Footnotes
The authors report no conflicts of interest.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.annalsofsurgery.com).
REFERENCES
- 1.Rahib L, Smith BD, Aizenberg R, et al. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and pancreas cancers in the United States. Cancer Res. 2014;74:2913–2921. [DOI] [PubMed] [Google Scholar]
- 2.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin. 2018;68:7–30. [DOI] [PubMed] [Google Scholar]
- 3.Kamisawa T,Wood LD,Itoi T,et al.Pancreatic cancer.Lancet.2016;388:73–85. [DOI] [PubMed] [Google Scholar]
- 4.Moaven O, Richman JS, Reddy S, et al. Healthcare disparities in outcomes of patients with resectable pancreatic cancer. Am J Surg. 2019;217:725–731. [DOI] [PubMed] [Google Scholar]
- 5.Neoptolemos JP, Kleeff J, Michl P, et al. Therapeutic developments in pancreatic cancer: current and future perspectives. Nat Rev Gastroenterol Hepatol. 2018;15:333–348. [DOI] [PubMed] [Google Scholar]
- 6.Strobel O, Neoptolemos J, Jager D, et al. Optimizing the outcomes of pancreatic cancer surgery. Nat Rev Clin Oncol. 2019;16:11–26. [DOI] [PubMed] [Google Scholar]
- 7.O’Reilly D, Fou L, Hasler E, et al. Diagnosis and management of pancreatic cancer in adults: a summary of guidelines from the UK National Institute for Health and Care Excellence. Pancreatology. 2018;18:962–970. [DOI] [PubMed] [Google Scholar]
- 8.Khorana AA, Mangu PB, Berlin J, et al. Potentially curable pancreatic cancer: american society of clinical oncology clinical practice guideline update. J Clin Oncol. 2017;35:2324–2328. [DOI] [PubMed] [Google Scholar]
- 9.Tempero MA, Malafa MP, Al-Hawary M, et al. Pancreatic adenocarcinoma,Version 2.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2017;15:1028–1061. [DOI] [PubMed] [Google Scholar]
- 10.Badiyan SN, Molitoris JK, Chuong MD, et al. The role of radiation therapy for pancreatic cancer in the adjuvant and neoadjuvant settings. Surg Oncol Clin N Am. 2017;26:431–453. [DOI] [PubMed] [Google Scholar]
- 11.Bilimoria KY, Stewart AK, Winchester DP, et al. The National Cancer DataBase: a powerful initiative to improve cancer care in the United States. Ann Surg Oncol. 2008;15:683–690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Oettle H, Neuhaus P, Hochhaus A, et al. Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial. JAMA. 2013; 310:1473–1481. [DOI] [PubMed] [Google Scholar]
- 13.Neoptolemos JP, Stocken DD, Friess H, et al. A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer. N Engl J Med. 2004;350:1200–1210. [DOI] [PubMed] [Google Scholar]
- 14.Kalser MH, Ellenberg SS. Pancreatic cancer. Adjuvant combined radiation and chemotherapy following curative resection. Arch Surg. 1985; 120:899–903. [DOI] [PubMed] [Google Scholar]
- 15.Gastrointestinal Tumor Study Group. Further evidence of effective adjuvant combined radiation and chemotherapy following curative resection of pancreatic cancer. Cancer. 1987;59:2006–2010. [DOI] [PubMed] [Google Scholar]
- 16.Klinkenbijl JH, Jeekel J, Sahmoud T, et al. Adjuvant radiotherapy and 5-fluorouracil after curative resection of cancer of the pancreas and periampullary region: phase III trial of the EORTC gastrointestinal tract cancer cooperative group. Ann Surg. 1999;230:776–782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Garofalo MC, Regine WF, Tan MT. On statistical reanalysis, the EORTC trialis a positive trial for adjuvant chemoradiation in pancreatic cancer. Ann Surg. 2006;244:332–333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Neoptolemos JP, Dunn JA, Stocken DD, et al. Adjuvant chemoradiotherapy and chemotherapy in resectable pancreatic cancer: a randomised controlled trial. Lancet. 2001;358:1576–1585. [DOI] [PubMed] [Google Scholar]
- 19.Hsu CC, Herman JM, Corsini MM, et al. Adjuvant chemoradiation for pancreatic adenocarcinoma: the Johns Hopkins Hospital-Mayo Clinic collaborative study. Ann Surg Oncol. 2010;17:981–990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rutter CE, Park HS, Corso CD, et al. Addition of radiotherapy to adjuvant chemotherapy is associated with improved overall survival in resected pancreatic adenocarcinoma: an analysis of the National Cancer Data Base. Cancer. 2015;121:4141–4149. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.