Abstract
Background:
Multimodal therapy, combining chemotherapy and surgical resection, is the standard of care for patients with resectable Pancreatic Ductal Adenocarcinoma (PDAC). Patients with poorly differentiated PDAC have suboptimal overall survival (OS), partly due to the limited understanding of optimal timing for return to intended oncologic therapy (RIOT) after upfront surgery. We aim to evaluate patients with poorly differentiated PDAC who undergo upfront surgical resection and assess how the timing of receipt of adjuvant chemotherapy is associated with OS.
Methods:
Using the National Cancer Database, we identified poorly differentiated non-metastatic PDAC patients who received upfront surgical resection followed by adjuvant chemotherapy (2007–2016). Adjusted Cox proportional hazard models evaluated OS based on RIOT timing. Logistic regression was used to identify factors associated with RIOT.
Results:
Of 2,737 included patients, 66.1% (n=1,810) did RIOT within 9 weeks and 33.9% (n=927) after 9 weeks. The median age of the study cohort was 62 years; 52% (n = 1,583) were male and 87.1% (n = 2,657) were White. Adjusted multivariable analysis noted that patients who did RIOT within 9 weeks were associated with improved OS compared to patients who did RIOT after 9 weeks (aHR 0.90, 95%CI 0.82–0.99, p=0.03). Treatment at academic facilities was associated with lower odds of RIOT within 9 weeks compared to treatment at non-academic facilities (aOR 0.83, 95%CI 0.70–0.98, p=0.03).
Conclusion:
Patients with poorly differentiated pancreatic ductal adenocarcinoma may benefit from adjuvant therapy within 9 weeks after surgical resection. Opportunities may exist in academic facilities to facilitate adjuvant therapy within 9 weeks of surgical resection.
Keywords: Pancreatic ductal adenocarcinoma, return to intended oncologic therapy, overall survival, poorly differentiated, National Cancer Database
Graphical Abstract
1. Background
Pancreatic Ductal Adenocarcinoma (PDAC) is the 3rd leading cause of cancer-related death in the United States after lung and colon cancer. In 2024, it was estimated that 66,440 people would be diagnosed with PDAC and 51,750 people would die of PDAC.1 It is a lethal disease and associated with five-year survival rate of 12.8%.2 This is due to several reasons: 1) lack of screening for prevention or early detection of PDAC3 with 80% newly diagnosed PDAC being non-operable;2 2) suboptimal chemotherapeutic regimens;4,5 3) significant weight loss and malnutrition related to this disease, results in decreased functionality and appetite for patients making it challenging to tolerate therapy to treat this disease.6,7
In patients with localized non-metastatic PDAC, multimodal therapy (i.e., surgical resection, chemotherapy or radiation therapy) is associated with improved overall survival (OS).8–10 The current standard of care for resectable or borderline-resectable disease includes a combination of chemotherapy and surgical resection (Figure 1).8 For resectable disease, neoadjuvant chemotherapy or upfront surgical resection are standard treatment options.7 Many academic centers prefer neoadjuvant chemotherapy in resectable PDAC patients to maximize the receipt of chemotherapy since 40% of patients are unable to complete adjuvant chemotherapy.4,9 This is important because patients with PDAC who undergo surgical resection alone have significantly lower median OS compared to patients who receive a combination of chemotherapy (either in neoadjuvant or adjuvant setting) and surgical resection.11 Due to the lack of randomized data (ongoing clinical trial),10 many centers opt for upfront surgical resection for patients with resectable PDAC followed by adjuvant chemotherapy.
Figure 1.
Stages of Pancreatic Ductal Adenocarcinoma (A) Resectable (B) Borderline-resectable (C) Locally advanced (D) Oligometastatic (E) Metastatic (adapted with permission from Springfield et al.).31
Despite multimodal therapy, the median OS is poor (< 2 years) in resectable PDAC.4,9,10,12 The factors associated with poor prognosis include poorly differentiated histology, lymphovascular invasion, perineural invasion, margin-positive resection, lymph node positive disease, size of the primary tumor and absence of adjuvant therapy.13 Limited data are available on patients with poorly differentiated resectable PDAC who undergo upfront surgical resection in relation to the optimal timing of adjuvant chemotherapy. The current National Comprehensive Cancer Network (NCCN) guidelines suggest initiating adjuvant chemotherapy ideally within 12 weeks, but after adequate postoperative recovery.8 When adjuvant chemotherapy was started within 12 weeks of surgical resection in patients with PDAC, no difference in OS was noted compared to patients who started adjuvant chemotherapy after 12 weeks.14,15 However, the study cohort in both retrospective studies was not limited to patients with poorly differentiated PDAC. Additionally, only 33.5% of PDAC patients have poorly differentiated histology making it challenging to study this population prospectively.16 Another study (n=7,548) analyzed the National Cancer Database (NCDB), where PDAC patients with Return to Intended Oncologic Therapy (RIOT) within 59 days (8.42 weeks) demonstrated an association with improvement in 2-year OS compared to PDAC patients with RIOT after 59 days (52.5% vs. 45.1%, p=0.02). In this study, 54% of patients had either well-differentiated or moderately differentiated PDAC, highlighting the need to evaluate the optimal timing of RIOT in a cohort of poorly differentiated PDAC patients.17 Hence, we analyzed the NCDB to identify patients with poorly differentiated PDAC who undergo upfront surgical resection and explore how the timing of RIOT is associated with OS.
2. Methods
2.1. Study Design
In this retrospective cohort study, the NCDB was used to identify patients with Pancreatic Ductal Adenocarcinoma (PDAC) from 2006 to 2017. The NCDB, started in 1989, is a joint initiative by the American College of Surgeons Commission on Cancer (CoC) and the American Cancer Society and is a nationwide oncology outcomes database. It contains hospital registry data, representing 72% of all newly diagnosed cancer cases in the United States with approximately 40 million records from more than 1500 CoC-accredited cancer programs. Data on all types of cancer are tracked, analyzed, and used to explore outcomes in cancer care and serve as the basis for quality improvement.18 The STROBE guidelines were followed during manuscript preparation.19
2.2. Study Patients
No institutional review board approval was required, as this is a retrospective study using deidentified data from the NCDB. We selected all patients diagnosed with PDAC using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) codes for PDAC (8020 and 8140). For primary site, the following codes were used: ‘C250’, ‘C251’, ‘C252’, ‘C253’, ‘C257’, ‘C258’, ‘C259’; not ‘C254’ (malignant neoplasm of the endocrine pancreas). Tumor grade/differentiation was obtained from the NCDB GRADE variable (2006–2017), which classifies tumors from well-differentiated (Grade I) to undifferentiated (Grade IV). Patients with poorly differentiated (Grade III) non-metastatic PDAC who received upfront surgical resection followed by adjuvant chemotherapy were included. Patients who received postoperative radiation were excluded from our study cohort as only patients who were coded as 0 on the Radiation/Surgery Sequence variable were included. The study period (2006–2017) was chosen to provide sufficient follow-up for OS. Patients who received neoadjuvant therapy prior to surgical resection were excluded using the variable “systemic/surgery sequence”. Patients with unknown data on the number of days between the date of initial diagnosis and the date of surgical resection and the number of days between the date of initial diagnosis and the start of systemic therapy were excluded. Patients who received only palliative care, those with metastatic disease on clinical staging, those who received upfront radiation and those with well or moderately differentiated PDAC were excluded from the final analysis cohort. See Figure 2.
Figure 2.
Schematic depicting inclusion and exclusion criteria of patients with pancreatic ductal adenocarcinoma (n=2,737) from the NCDB.
Abbreviations: NCDB, National Cancer Database; RIOT, Return to Intended Oncologic Therapy
2.3. Variables and Comparison Groups
Independent variables in this analysis included patient demographic and pathological tumor factors. Patient demographic factors included age, race (white, black and others), facility type (academic/research vs. non-academic) and insurance status (private vs. Medicare/Medicaid/other government/not insured); these were treated as unordered categorical variables. Neighborhood-level education and median household income were defined using ZIP code–level estimates. Education was classified into quartiles based on the percentage of adults in the patient’s ZIP code without a high school diploma (<7%, 7% to 12.9%, 13% to 20.9%, and ≥21%). Median household income was categorized into quartiles according to ZIP code (<$38,000, $38,000 to $47,999, $48,000 to $67,999, and ≥$68,000). The Charlson Deyo (CD) score (0 vs. ≥1) was treated as an ordered categorical variable. Pathological tumor features included margin status (positive vs. negative) which was treated as an unordered categorical variable and pathological T and M staging which were treated as ordered categorical variables. Timing to RIOT was treated as an ordered categorical variable. The year of diagnosis was defined as the year when cancer was clinically or histologically confirmed, based on the Facility Oncology Registry Data Standards (FORDS) definition. Time interval was defined as time between definitive surgical procedure and start of adjuvant chemotherapy. Patients were right-censored at the time of death or last known follow-up, with survival time calculated from the date of diagnosis. Dichotomizing the RIOT time interval at 5, 6, 7, 8, 9, 10, 11, 12 and 16 weeks was evaluated to assess the impact on OS.
2.4. Statistical Analysis
Descriptive statistics were used to summarize patient demographics and tumor characteristics. Exploratory analyses were conducted at multiple RIOT time points (5, 6, 7, 8, 9, 10, 11, 12, and 16 weeks) to identify the threshold with a statistically significant association with OS in multivariable Cox models. In the multivariable Cox proportional hazard model, patient characteristics (age, sex, race, facility type, insurance status, CD score, median household income, education and year of diagnosis) and tumor characteristics (margin status and pathological T and N staging) were initially considered for model inclusion. Then a backward selection method was used for multivariable analysis with an alpha level of removal of >/=0.2. To identify factors associated with RIOT, a binary logistic regression model was used to assess patient demographics and tumor factors associated with RIOT within the optimal time interval based on our exploratory analyses. A landmark analysis was conducted with a landmark time of 180 days post-surgery, excluding patients who died within 180 days of surgical resection. Missing data were handled using a complete case analysis approach in the multivariable models. A p<0.05 was accepted to indicate a statistically significant association. All analyses were performed using SAS 9.4 (SAS Institute Inc.).
3. Results
3.1. Patient Demographics and Tumor Characteristics
We reviewed de-identified data for 585,239 patients with pancreatic cancer from the NCDB between 2006 and 2017. Following the application of selection criteria (Figure 2), 2,737 patients were included in the final analysis. The cohort predominantly consisted of White patients (n=2,657, 87.1%), with 65.4% having a CD score of 0 (n=1,995). A majority of patients (n= 2,124, 69.6%) were diagnosed between 2011 and 2017. Regarding insurance status, 38.7% (n=1,181) had private insurance, while the remaining 61.3% (n=1,869) were either insured through Medicare/Medicaid/other government programs or uninsured. Advanced tumor stage (pT3–4) was observed in 85.4% (n=2,441) patients, with nodal involvement (pN1) in 72.6% (n = 2,059) patients. Negative surgical margins were reported in 77.9% (n=2,375) patients. The median time from diagnosis to surgery was 2.4 (IQR 0.4–4.1) weeks (Table 1). The median time from surgery to adjuvant therapy was 7.8 (IQR 6–10) weeks. On a median follow-up of 13 (IQR 9.6–18.4) weeks, 87% of patients (n=2,653) died, while the remaining 13% (n=397) were right-censored (alive or lost to follow-up).
Table 1.
Descriptive Statistics
| Variable | Level | N (%) (N=2,737) |
|---|---|---|
| Race | White | 2394 (87.5) |
| Black | 230 (8.4) | |
| Other | 113 (4.1) | |
| Age | Median | 62 years |
| Below Median | 1440 (52.6) | |
| Above Median | 1297 (47.4) | |
| Sex | Male | 1417 (51.8) |
| Female | 1320 (48.2) | |
| Year of Diagnosis | 2006–2010 | 801 (29.3) |
| 2011–2017 | 1936 (70.7) | |
| Charlson-Deyo Score | 0 | 1815 (66.3) |
| 1+ | 922 (33.7) | |
| Insurance Status | Private | 1072 (39.2) |
| Medicare/Medicaid/Other Government/Not Insured | 1665 (60.8) | |
| Pathologic T | 1–2 | 396 (15.3) |
| 3–4 | 2187 (84.7) | |
| Missing | 154 | |
| Pathologic N | 0 | 720 (28.1) |
| 1 | 1841 (71.9) | |
| Missing | 176 | |
| Median Household Income | >=$68,000 | 922 (37.0) |
| $48,000–$67,999 | 671 (27.0) | |
| $38,000–$47,999 | 546 (21.9) | |
| <$38,000 | 350 (14.1) | |
| Missing | 248 | |
| Education (% No High School Degrees Quartile) | <7.0% | 708 (28.4) |
| 7.0–12.9% | 808 (32.4) | |
| 13.0–20.9% | 622 (25.0) | |
| >=21% | 352 (14.1) | |
| Missing | 247 | |
| Surgical Margins | Negative | 2190 (81.7) |
| Positive | 489 (18.3) | |
| Missing | 58 | |
| RIOT within 9 weeks | Yes | 1810 (66.1) |
| No | 927 (33.9) | |
| Time from diagnosis to surgery (days) | Mean | 21.01 |
| Median | 17.00 | |
| Range | 0–409 | |
| Interquartile Range | 3–29 | |
| Time from surgery to adjuvant therapy | Mean | 59.14 |
| Median | 55.00 | |
| Range | 0–417 | |
| Interquartile Range | 42–70 |
RIOT, Return to Intended Oncologic Therapy
3.2. Demographic and Tumor Characteristics associated with Overall Survival
On univariate analysis, RIOT within 9 weeks, age, facility type, insurance status, CD score, year of diagnosis, pathologic T and N staging and margin status were associated with improved OS (Table 2). On univariate analysis, RIOT within 8 weeks (p=0.016) and 9 weeks (p=0.025) were associated with improved OS (Table 2), while other time points (5, 6, 7, 10, 11, 12, and 16 weeks) showed no significant association (Supplementary Table 2). On multivariable analysis, RIOT within 9 weeks remained significantly associated with improved OS (aHR 0.90, 95%CI 0.82–0.99, p=0.03). Similar results were observed in two other multivariable models: 1) Patients who did RIOT within 8 weeks vs. those who did RIOT after 8 weeks (aHR 0.89, 95%CI 0.82–0.98, p=0.015) and 2) Patients who did RIOT within 7 weeks vs. those who did RIOT after 7 weeks (aHR 0.9, 95%CI 0.82–0.99, p=0.029). No association with OS was observed in multivariable models comparing patients who did RIOT within 5, 6, 10, 11, 12 and 16 weeks vs. those who did RIOT after these time points following upfront surgery (Supplementary Table 3). A landmark analysis at 180 days post-surgery confirmed the association between improved OS with RIOT within 8 (aHR 0.90, 95% CI 0.83–0.98, p=0.018) and 9 weeks (aHR 0.91, 95% CI 0.83–0.98, p=0.026), supporting the primary findings while mitigating immortal time bias. Other factors on multivariable analysis associated with improved OS included treatment at academic/research facility compared to the non-academic facility (aHR 0.83, 95%CI 0.76–0.91, p<0.001), year of diagnosis between 2011 and 2017 compared to 2006–2011 (aHR 0.84, 95%CI 0.76–0.93, p=0.001), pathologic T1–2 stage compared to T3–4 stage (aHR 0.75, 95%CI 0.66–0.86, p<0.001), pathologic N0 stage compared to N1 stage (aHR 0.71, 95%CI 0.64–0.79, p<0.001) and negative surgical margins compared to positive margins (aHR 0.69, 95%CI 0.62–0.78, p<0.001). Factors associated with worse OS included Black race compared to White (aHR 1.29, 95%CI 1.1–1.52, p=0.002), and a CD score of 1 compared to 0 (aHR 1.12, 95%CI 1.02–1.24, p=0.016) (Table 3).
Table 2.
Factors associated with overall survival based on univariate Cox regression analysis
| Overall Survival | ||||||
|---|---|---|---|---|---|---|
| Covariate | Level | N | Hazard Ratio (95% CI) | HR P-value | Log-rank P-value | |
| RIOT within 8 weeks | Yes | 1503 | 0.90 (0.83–0.98) | 0.016 | 0.016 | |
| No | 1234 | - | - | |||
| RIOT within 9 weeks | Yes | 1810 | 0.91 (0.83–0.99) | 0.025 | 0.025 | |
| No | 927 | - | - | |||
| Age | Above Median | 1297 | 1.14 (1.05–1.23) | 0.002 | 0.002 | |
| Below Median | 1440 | - | - | |||
| Race | Black | 230 | 1.22 (1.06–1.41) | 0.007 | 0.017 | |
| Other | 113 | 0.93 (0.75–1.14) | 0.480 | |||
| White | 2394 | - | - | |||
| Sex | Female | 1320 | 0.95 (0.87–1.03) | 0.187 | 0.186 | |
| Male | 1417 | - | - | |||
| Facility Type | Academic/Research Program | 1361 | 0.85 (0.78–0.92) | <.001 | <.001 | |
| Non-Academic | 1349 | - | - | |||
| Insurance Status | Private | 1072 | 0.89 (0.82–0.97) | 0.007 | 0.007 | |
| Medicare/Medicaid/Other Government/Not Insured | 1665 | - | - | |||
| Charlson-Deyo Score | 1+ | 922 | 1.13 (1.04–1.23) | 0.005 | 0.005 | |
| 0 | 1815 | - | - | |||
| Year of Diagnosis | 2011–2017 | 1936 | 0.84 (0.77–0.92) | <.001 | <.001 | |
| 2006–2010 | 801 | - | - | |||
| Pathologic T | 1–2 | 396 | 0.71 (0.63–0.80) | <.001 | <.001 | |
| 3–4 | 2187 | - | - | |||
| Pathologic N | 0 | 720 | 0.66 (0.60–0.72) | <.001 | <.001 | |
| 1 | 1841 | - | - | |||
| Margins | Negative | 2190 | 0.63 (0.57–0.70) | <.001 | <.001 | |
| Positive | 489 | - | - | |||
| Median Household Income | >=$68,000 | 922 | 0.88 (0.77–1.00) | 0.059 | 0.225 | |
| $48,000–$67,999 | 671 | 0.88 (0.77–1.02) | 0.083 | |||
| $38,000–$47,999 | 546 | 0.87 (0.75–1.01) | 0.064 | |||
| <$38,000 | 350 | - | - | |||
| Education (% No High School Degrees Quartile) | <7.0% | 708 | 1.01 (0.88–1.16) | 0.912 | 0.100 | |
| 7.0–12.9% | 808 | 0.97 (0.84–1.11) | 0.615 | |||
| 13.0–20.9% | 622 | 1.11 (0.96–1.28) | 0.148 | |||
| >=21% | 352 | - | - | |||
RIOT, Return to Intended Oncologic Therapy
Results for other RIOT time points (5, 6, 7, 10, 11, 12, and 16 weeks) are non-significant and reported in Supplementary Table 2.
Table 3.
Factors associated with overall survival based on multivariable Cox regression analysis
| Overall Survival | ||||
|---|---|---|---|---|
| Covariate | Level | Hazard Ratio (95% CI) | HR P-value | Overall P-value |
| RIOT within 9 weeks | Yes | 0.90 (0.82–0.99) | 0.032 | 0.032 |
| No | - | - | ||
| Age | Above Median | 1.09 (0.98–1.21) | 0.113 | 0.113 |
| Below Median | - | - | ||
| Race | Black | 1.29 (1.10–1.52) | 0.002 | 0.006 |
| Other | 0.92 (0.72–1.16) | 0.473 | ||
| White | - | - | ||
| Facility Type | Academic/Research Program | 0.83 (0.76–0.91) | <.001 | <.001 |
| Non-Academic | - | - | ||
| Insurance Status | Private | 0.91 (0.82–1.02) | 0.103 | 0.103 |
| Medicare/Medicaid/Other Government/Not Insured | - | - | ||
| Charlson-Deyo Score | 1+ | 1.12 (1.02–1.24) | 0.016 | 0.016 |
| 0 | - | - | ||
| Year of Diagnosis | 2011–2017 | 0.84 (0.76–0.93) | 0.001 | 0.001 |
| 2006–2010 | - | - | ||
| Pathologic T | 1–2 | 0.75 (0.66–0.86) | <.001 | <.001 |
| 3–4 | - | - | ||
| Pathologic N | 0 | 0.71 (0.64–0.79) | <.001 | <.001 |
| 1 | - | - | ||
| Margins | Negative | 0.69 (0.62–0.78) | <.001 | <.001 |
| Positive | - | - | ||
| Education (% No High School Degrees Quartile) | <7.0% | 1.05 (0.90–1.22) | 0.567 | 0.096 |
| 7.0–12.9% | 0.97 (0.83–1.12) | 0.656 | ||
| 13.0–20.9% | 1.12 (0.96–1.31) | 0.140 | ||
| >=21% | - | - | ||
Number of observations in the original data set = 2737. Number of observations used = 2260.
Backward selection with an alpha level of removal of .2 was used. The following variables were removed from the model: Median Household Income.
3.3. Demographic and Tumor Characteristics associated with RIOT (≤ 9 weeks)
Following surgical resection, 66.1% (n=1,810) of patients were able to RIOT within 9 weeks, while 33.9% (n=927) patients were able to RIOT after 9 weeks (Table 1). On univariate analysis, age below the median (OR 1.21, 95%CI 1.04–1.42, p=0.016), median household income quartile ($48,000-$67,999: OR 1.21, 95%CI 1.04–1.42, p=0.016; $38,000-$47,999: OR 1.21, 95%CI 1.04–1.42, p=0.016; Ref. ≥$68,000) and private insurance (OR 1.25, 95%CI 1.06–1.47, p=0.008) were associated with higher odds of RIOT within 9 weeks (Table 4). On multivariable analysis, treatment at a non-academic facility was associated with higher odds of RIOT within 9 weeks compared to treatment at an academic/research facility (aOR 1.18, 95%CI 1.01–1.38, p=0.03) (Table 5). Patients with a median household income of $48,000–$67,999 (aOR 0.72, 95%CI 0.56–0.91, p=0.007) and $38,000–$47,999 (aOR 0.66, 95%CI 0.50–0.86, p=0.003) were associated with lower odds of RIOT within 9 weeks compared to those in the highest income quartile (≥$68,000). The odds of RIOT within 9 weeks were not significantly different for patients in the lowest income quartile (<$38,000) compared to those in the highest income quartile (≥$68,000) (aOR 0.86, 95%CI 0.61–1.21, p=0.383).
Table 4.
Factors associated with RIOT within 9 weeks based on univariate logistic regression analysis
| RIOT within 9 weeks | |||||
|---|---|---|---|---|---|
| Covariate | Level | N | Odds Ratio (95% CI) | OR P-value | Overall P-value |
| Age | Below Median | 1440 | 1.21 (1.04–1.42) | 0.016 | 0.016 |
| Above Median | 1297 | - | - | ||
| Race | White | 2394 | 1.09 (0.74–1.62) | 0.656 | 0.308 |
| Black | 230 | 0.88 (0.55–1.41) | 0.606 | ||
| Other | 113 | - | - | ||
| Facility Type | Non-Academic | 1349 | 1.17 (1.00–1.37) | 0.057 | 0.057 |
| Academic/Research Program | 1361 | - | - | ||
| Insurance Status | Private | 1072 | 1.25 (1.06–1.47) | 0.008 | 0.008 |
| Medicare/Medicaid/Other Government/Not Insured | 1665 | - | - | ||
| Charlson-Deyo Score | 0 | 1815 | 1.15 (0.98–1.36) | 0.092 | 0.092 |
| 1+ | 922 | - | - | ||
| Year of Diagnosis | 2006–2010 | 801 | 0.99 (0.83–1.17) | 0.879 | 0.879 |
| 2011–2017 | 1936 | - | - | ||
| Pathologic T | 1–2 | 396 | 0.93 (0.74–1.16) | 0.522 | 0.522 |
| 3–4 | 2187 | - | - | ||
| Pathologic N | 0 | 720 | 0.98 (0.81–1.17) | 0.803 | 0.803 |
| 1 | 1841 | - | - | ||
| Margins | Negative | 2190 | 1.06 (0.87–1.31) | 0.559 | 0.559 |
| Positive | 489 | - | - | ||
| Median Household Income | $48,000–$67,999 | 671 | 0.77 (0.62–0.95) | 0.014 | 0.012 |
| $38,000–$47,999 | 546 | 0.71 (0.57–0.89) | 0.002 | ||
| <$38,000 | 350 | 0.83 (0.64–1.07) | 0.150 | ||
| >=$68,000 | 922 | - | - | ||
| Education (% No High School Degrees Quartile) | <7.0% | 708 | 1.29 (0.99–1.68) | 0.059 | 0.242 |
| 7.0–12.9% | 808 | 1.28 (0.99–1.66) | 0.065 | ||
| 13.0–20.9% | 622 | 1.23 (0.94–1.62) | 0.131 | ||
| >=21% | 352 | - | - | ||
RIOT, Return to Intended Oncologic Therapy
Table 5.
Factors associated with RIOT within 9 weeks based on multivariable logistic regression analysis
| RIOT within 9 weeks | ||||
|---|---|---|---|---|
| Covariate | Level | N | Odds Ratio (95% CI) | OR P-value |
| Age | Above Median | 1182 | 0.87 (0.71–1.05) | 0.152 |
| Below Median | 1284 | - | - | |
| Facility Type | Non-Academic | 1226 | 1.18 (1.01–1.38) | 0.030 |
| Academic/Research Program | 1240 | - | - | |
| Insurance Status | Private | 957 | 1.16 (0.95–1.42) | 0.148 |
| Medicare/Medicaid/Other Government/Not Insured | 1509 | - | - | |
| Charlson-Deyo Score | 1+ | 829 | 0.85 (0.71–1.01) | 0.063 |
| 0 | 1637 | - | - | |
| Median Household Income | $48,000–$67,999 | 666 | 0.72 (0.56–0.91) | 0.007 |
| $38,000–$47,999 | 540 | 0.66 (0.50–0.86) | 0.003 | |
| <$38,000 | 345 | 0.86 (0.61–1.21) | 0.383 | |
| >=$68,000 | 915 | - | - | |
| Education (% No High School Degrees Quartile) | <7.0% | 702 | 1.08 (0.77–1.53) | 0.645 |
| 7.0–12.9% | 804 | 1.29 (0.96–1.75) | 0.095 | |
| 13.0–20.9% | 612 | 1.29 (0.97–1.72) | 0.078 | |
| >=21% | 348 | - | - | |
Number of observations in the original data set = 2737. Number of observations used = 2466.
Backward selection with an alpha level of removal of .2 was used. The following variables were removed from the model: Margins, AJCC Pathologic N, AJCC Pathologic T, Year of Diagnosis, and Race.
4. Discussion
The primary aim of this study was to evaluate the optimal timing of RIOT and its association with OS in patients with poorly differentiated PDAC who underwent upfront surgical resection. The secondary aim was to evaluate factors associated with timely RIOT in poorly differentiated PDAC. The analysis determined that 9 weeks is the optimal time cut-off for RIOT, before which patients may benefit from initiating adjuvant therapy. Patients with RIOT within 9 weeks demonstrated an association with improved OS compared to patients with RIOT after 9 weeks. Although RIOT within 7 or 8 weeks was also associated with improved OS, we selected the 9-week cut-off as the primary result because it provided the most clinically meaningful threshold to start adjuvant therapy. In order to decrease the risk of bias, we analyzed the data with landmark analysis, which supported our primary findings. Treatment at a non-academic facility was associated with higher odds of RIOT within 9 weeks compared to treatment at an academic/research facility.
Several retrospective studies analyzed the association between timing of RIOT and OS.14,15,20 Analysis of patients with stage 1 and 2 PDAC (n=7,548) noted that patients who had RIOT before 28 days (HR 1.17, 95%CI 1.02–1.35, p=0.03) and after 59 days (HR 1.09, 95%CI 1.02–1.17, p=0.008) had worse OS compared to patients who had RIOT between 28 and 59 days following surgical resection.20 In another multi-institutional study using the central pancreatic consortium database (n=488), similar OS was noted in PDAC patients with RIOT within 12 weeks compared to patients with RIOT after 12 weeks on univariate analysis (24.3 vs. 28.5 months, p = 0.79).14 On multivariable analysis, patients with RIOT, regardless of timing, were noted to have improved OS compared to patients who did not RIOT (≤12 weeks: HR 0.5, 95%CI 0.34–0.82, p<0.01; >12 weeks: HR 0.48, 95%CI 0.29–0.8, p<0.01; reference: surgery only). The authors concluded that patients who did not RIOT within 12 weeks of surgical resection still remained appropriate candidates for multimodal therapy. Unlike these two studies, we limited our analysis to PDAC patients with poorly differentiated histology and noted that RIOT within 9 weeks was associated with improved OS. In another study, poorly differentiated histology was identified as an independent predictor of failure to RIOT (OR 0.62) in resected PDAC patients, underscoring the need to optimize RIOT in this population to enhance oncologic outcomes. Our study offers an additional framework to guide the optimal timing of RIOT.21
Poorly differentiated PDAC is associated with early recurrence and worse OS.22,23 In a single US-based institutional analysis of patients with resected PDAC (n=957), higher odds of early recurrence were noted in patients with poorly differentiated tumors on surgical pathology compared to well/moderately differentiated tumors (OR 1.66, 95%CI 1.1–2.5, p=0.016).24 In another single-center analysis of patients with resected PDAC (n=510), disease-free survival was compared in patients with RIOT compared to patients without RIOT.16 In a subgroup analysis of patients with poorly differentiated PDAC, a significant improvement in median disease-free survival was associated with RIOT compared to patients without RIOT (22 vs. 12 months, p<0.0001). This association was not significant in a subgroup of patients with well-differentiated PDAC (p=0.25). Our study focused on OS as the main outcome as it is clinically meaningful due to the overall prognosis of PDAC. We noted that the timing of RIOT is also an important prognostic factor in these patients due to its association with OS. Improved OS was noted in patients with poorly differentiated PDAC with RIOT within 9 weeks compared to RIOT after 9 weeks. Recent studies highlight the increasing use of neoadjuvant therapy for PDAC.25 Our study cohort consisted of patients with upfront surgical resection, which may limit the broad applicability of our findings to patients receiving neoadjuvant therapy. These data highlight the importance of RIOT and timing of RIOT in patients with poorly differentiated PDAC.
Several studies have identified factors associated with RIOT in patients with PDAC.27–29 In an NCDB analysis (2004–2016) by Mickel et al., male sex compared to female (OR 1.24, p<0.001), CD score of 0 (OR 1.9, p<0.001) and 1 (OR 1.7, p=0.001) compared to ≥2, negative surgical margins compared to positive margins (OR 1.1, p=0.048) and treatment at academic institutions compared to community hospital (OR 1.3, p=0.001) were associated with improved odds of RIOT.29 In contrast to this study, we assessed factors associated with RIOT within 9 weeks in poorly differentiated PDAC and noted that patients receiving treatment at non-academic facilities were associated with higher odds of RIOT within 9 weeks compared to patients receiving treatment at academic facilities. One potential barrier to obtaining treatment in a timely fashion at an academic center may be delays in referrals from a non-academic center, or delays in obtaining appointments for further care, including adjuvant therapy.30 Distance to treating facility, or outcomes based on rural vs. other geographic areas of residency were not included in the analyses. In addition, examining access to care based on having private insurance vs. government or no insurance at all may not provide sufficient nuance to identify how or if payor type might impact timely receipt of RIOT. Our analysis noted an improved OS in patients receiving treatment at academic facilities compared to non-academic facilities. Our study highlights the complex interaction between the treatment facility, RIOT and OS in patients with poorly differentiated PDAC. Future studies should aim at identifying factors associated with RIOT within the optimal timeframe from surgical resection in patients with poorly differentiated PDAC to understand the generalizability of these findings.
The limitations of our study include its retrospective nature and the associated selection bias, which may have influenced RIOT rates and the timing of RIOT. Since the NCDB is a hospital-based database, it includes only patients treated at Commission on Cancer-accredited facilities and is not representative of the entire U.S. population. This may limit the generalizability of our findings. We did not evaluate postoperative complications, as NCDB does not record specific postoperative complication data, which may be a critical factor in timely administration of RIOT. We acknowledge the lack of details in the NCDB related to adjuvant therapy, such as the name, dose, duration, frequency and toxicity / adverse events. Due to this limitation, we were unable to determine which patients actually completed the intended adjuvant therapy. The NCDB does not provide data on disease-free survival or recurrence, limiting our ability to assess the impact of RIOT timing on these outcomes. This is a known limitation in NCDB, as recurrence is a critical endpoint in PDAC. Unmeasured potential confounders, such as postoperative complications, patient performance status, and chemotherapy compliance, are not captured in the NCDB and may influence the association between the timing of RIOT and OS. This may limit the generalizability of our findings. Our survival analysis may also be confounded by unobservable differences in surgeon expertise, experience and the lack of details regarding annual surgical volume. Limiting our analysis to patients who received upfront resection may reduce the generalizability of our findings as many patients now receive neoadjuvant chemotherapy. Despite these limitations, our study investigates the impact of receipt and timing of RIOT on OS and identifies factors associated with RIOT within 9 weeks in patients with poorly differentiated PDAC.
In conclusion, patients with poorly differentiated pancreatic ductal adenocarcinoma who return to intended oncologic therapy within 9 weeks demonstrate an association with improved overall survival compared to patients who return to intended oncologic therapy after 9 weeks. Treatment at non-academic facility is associated with timely return to intended oncologic therapy within 9 weeks; the reason for this finding may be interesting to evaluate in future studies. Additional studies should aim at exploring other factors that may influence the timing of return to intended oncologic therapy to optimize the delivery of multimodal treatment and improve overall survival, especially for patients with poorly differentiated pancreatic ductal adenocarcinoma.
Supplementary Material
Funding Information:
Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health (NIH/NCI) under award number K12 CA237806 from the Emory K12 Clinical Oncology Training Program (awarded to MMS), the Georgia CTSA ULI TR002378 (awarded to MMS), and NIH/NCI P30CA138292 (awarded to Biostatistics Shared Resource). The Content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The acknowledged parties had no role in study design, data collection, analysis and interpretation of data, manuscript writing and decision to submit the manuscript for publication.
Footnotes
Disclosure:
This manuscript is part of a master’s thesis submitted to Emory University Laney Graduate School. The thesis is currently under embargo and is not publicly accessible. The content has not been previously published in any peer-reviewed journal.
References
- 1.Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA: A Cancer Journal for Clinicians. 2024;74(1):12–49. doi: 10.3322/caac.21820 [DOI] [PubMed] [Google Scholar]
- 2.Cancer of the Pancreas - Cancer Stat Facts. SEER. Accessed September 15, 2023. https://seer.cancer.gov/statfacts/html/pancreas.html [Google Scholar]
- 3.Singhi AD, Koay EJ, Chari ST, Maitra A. Early Detection of Pancreatic Cancer: Opportunities and Challenges. Gastroenterology. 2019;156(7):2024–2040. doi: 10.1053/j.gastro.2019.01.259 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sohal DPS, Duong M, Ahmad SA, et al. Efficacy of Perioperative Chemotherapy for Resectable Pancreatic Adenocarcinoma: A Phase 2 Randomized Clinical Trial. JAMA Oncology. 2021;7(3):421–427. doi: 10.1001/jamaoncol.2020.7328 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Neoptolemos JP, Stocken DD, Bassi C, et al. Adjuvant Chemotherapy With Fluorouracil Plus Folinic Acid vs Gemcitabine Following Pancreatic Cancer Resection: A Randomized Controlled Trial. JAMA. 2010;304(10):1073–1081. doi: 10.1001/jama.2010.1275 [DOI] [PubMed] [Google Scholar]
- 6.Bachmann J, Büchler MW, Friess H, Martignoni ME. Cachexia in patients with chronic pancreatitis and pancreatic cancer: impact on survival and outcome. Nutr Cancer. 2013;65(6):827–833. doi: 10.1080/01635581.2013.804580 [DOI] [PubMed] [Google Scholar]
- 7.Bachmann J, Heiligensetzer M, Krakowski-Roosen H, Büchler MW, Friess H, Martignoni ME. Cachexia worsens prognosis in patients with resectable pancreatic cancer. J Gastrointest Surg. 2008;12(7):1193–1201. doi: 10.1007/s11605-008-0505-z [DOI] [PubMed] [Google Scholar]
- 8.[Guideline] National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology. Pancreatic Adenocarcinoma. NCCN; Available at https://www.nccn.org/professionals/physician_gls/pdf/pancreatic.pdf. Version 3.2024 — August 2, 2024; Accessed: October 1, 2024. [Google Scholar]
- 9.Ahmad SA, Duong M, Sohal DPS, et al. Surgical Outcome Results From SWOG S1505: A Randomized Clinical Trial of mFOLFIRINOX Versus Gemcitabine/Nab-paclitaxel for Perioperative Treatment of Resectable Pancreatic Ductal Adenocarcinoma. Annals of Surgery. 2020;272(3):481–486. doi: 10.1097/SLA.0000000000004155 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.van Dam JL, Verkolf EMM, Dekker EN, et al. Perioperative or adjuvant mFOLFIRINOX for resectable pancreatic cancer (PREOPANC-3): study protocol for a multicenter randomized controlled trial. BMC Cancer. 2023;23(1):728. doi: 10.1186/s12885-023-11141-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.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(14):1473–1481. doi: 10.1001/jama.2013.279201 [DOI] [PubMed] [Google Scholar]
- 12.Labori KJ, Bratlie SO, Andersson B, et al. Neoadjuvant FOLFIRINOX versus Upfront Surgery for Resectable Pancreatic Head Cancer (NORPACT-1): A Multicentre, Randomised, Phase 2 Trial. The Lancet Gastroenterology & Hepatology. 2024;0(0). doi: 10.1016/S2468-1253(23)00405-3 [DOI] [PubMed] [Google Scholar]
- 13.Javed AA, Mahmud O, Fatimi AS, et al. Predictors for Long-Term Survival After Resection of Pancreatic Ductal Adenocarcinoma: A Systematic Review and Meta-Analysis. Ann Surg Oncol. 2024;31(7):4673–4687. doi: 10.1245/s10434-024-15281-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Xia BT, Ahmad SA, Al Humaidi AH, et al. Time to Initiation of Adjuvant Chemotherapy in Pancreas Cancer: A Multi-Institutional Experience. Ann Surg Oncol. 2017;24(9):2770–2776. doi: 10.1245/s10434-017-5918-z [DOI] [PubMed] [Google Scholar]
- 15.Ka M, Ek G, Cs H, J W. Time to the initiation of adjuvant chemotherapy does not impact survival in patients with resected pancreatic cancer. Cancer. 2016;122(19). doi: 10.1002/cncr.30163 [DOI] [PubMed] [Google Scholar]
- 16.Crippa S, Partelli S, Zamboni G, et al. Poorly differentiated resectable pancreatic cancer: Is upfront resection worthwhile? Surgery. 2012;152(3, Supplement):S112–S119. doi: 10.1016/j.surg.2012.05.017 [DOI] [PubMed] [Google Scholar]
- 17.Ma SJ, Oladeru OT, Miccio JA, Iovoli AJ, Hermann GM, Singh AK. Association of Timing of Adjuvant Therapy With Survival in Patients With Resected Stage I to II Pancreatic Cancer. JAMA Network Open. 2019;2(8):e199126. doi: 10.1001/jamanetworkopen.2019.9126 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.National Cancer Database. ACS. Accessed February 21, 2024. https://www.facs.org/quality-programs/cancer-programs/national-cancer-database/ [Google Scholar]
- 19.von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 20.Mirkin KA, Greenleaf EK, Hollenbeak CS, Wong J. Time to the initiation of adjuvant chemotherapy does not impact survival in patients with resected pancreatic cancer. Cancer. 2016;122(19):2979–2987. doi: 10.1002/cncr.30163 [DOI] [PubMed] [Google Scholar]
- 21.Mackay TM, Smits FJ, Roos D, et al. The risk of not receiving adjuvant chemotherapy after resection of pancreatic ductal adenocarcinoma: a nationwide analysis. HPB. 2020;22(2):233–240. doi: 10.1016/j.hpb.2019.06.019 [DOI] [PubMed] [Google Scholar]
- 22.Lim JE, Chien MW, Earle CC. Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based, linked database analysis of 396 patients. Ann Surg. 2003;237(1):74–85. doi: 10.1097/00000658-200301000-00011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cleary SP, Gryfe R, Guindi M, et al. Prognostic factors in resected pancreatic adenocarcinoma: analysis of actual 5-year survivors. J Am Coll Surg. 2004;198(5):722–731. doi: 10.1016/j.jamcollsurg.2004.01.008 [DOI] [PubMed] [Google Scholar]
- 24.Groot VP, Gemenetzis G, Blair AB, et al. Defining and Predicting Early Recurrence in 957 Patients With Resected Pancreatic Ductal Adenocarcinoma. Ann Surg. 2019;269(6):1154–1162. doi: 10.1097/SLA.0000000000002734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Davis CH, Augustinus S, de Graaf N, et al. Impact of Neoadjuvant Therapy for Pancreatic Cancer: Transatlantic Trend and Postoperative Outcomes Analysis. Journal of the American College of Surgeons. 2024;238(4):613. doi: 10.1097/XCS.0000000000000971 [DOI] [PubMed] [Google Scholar]
- 26.Wehrle CJ, Chang JH, Gross AR, et al. Comparing oncologic and surgical outcomes of robotic and laparoscopic pancreatoduodenectomy in patients with pancreatic cancer: a propensity-matched analysis. Surg Endosc. 2024;38(5):2602–2610. doi: 10.1007/s00464-024-10783-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lu PW, Lyu HG, Prakash LR, et al. Effect of surgical approach on early return to intended oncologic therapy after resection for pancreatic ductal adenocarcinoma. Surg Endosc. 2024;38(9):4986–4995. doi: 10.1007/s00464-024-11022-3 [DOI] [PubMed] [Google Scholar]
- 28.Naffouje SA, Kamarajah SK, Denbo JW, Salti GI, Dahdaleh FS. Surgical Approach does not Affect Return to Intended Oncologic Therapy Following Pancreaticoduodenectomy for Pancreatic Adenocarcinoma: A Propensity-Matched Study. Ann Surg Oncol. 2022;29(12):7793–7803. doi: 10.1245/s10434-022-12347-w [DOI] [PubMed] [Google Scholar]
- 29.Mickel TA, Kutlu OC, Silberfein EJ, et al. Factors associated with inability to return to intended oncologic treatment in pancreatic cancer. The American Journal of Surgery. 2022;224(1, Part B):635–640. doi: 10.1016/j.amjsurg.2022.02.058 [DOI] [PubMed] [Google Scholar]
- 30.Graboyes EM, Sterba KR, Li H, et al. Development and Evaluation of a Navigation-Based, Multilevel Intervention to Improve the Delivery of Timely, Guideline-Adherent Adjuvant Therapy for Patients With Head and Neck Cancer. JCO Oncol Pract. 2021;17(10):e1512–e1523. doi: 10.1200/OP.20.00943 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Springfeld C, Ferrone CR, Katz MHG, et al. Neoadjuvant therapy for pancreatic cancer. Nat Rev Clin Oncol. 2023;20(5):318–337. doi: 10.1038/s41571-023-00746-1 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.