Abstract
Background
Curative-intent treatment for pancreatic ductal adenocarcinoma necessitates major surgery and chemotherapy, yet many patients do not undergo resection. Frailty may contribute to this phenomenon, but its prevalence and association with pancreatectomy in pancreatic ductal adenocarcinoma remains poorly defined. This study describes the distribution of frailty and its association with curative-intent surgery in a nationally representative cohort of patients with nonmetastatic pancreatic ductal adenocarcinoma.
Methods
Using Surveillance, Epidemiology, and End Results–Medicare data (2013-2019), we identified patients aged 66 years and older with nonmetastatic pancreatic ductal adenocarcinoma. Frailty was categorized as nonfrail, prefrail, mildly frail, or moderately to severely frail using the claims-based frailty index over the 12 months preceding diagnosis. We evaluated frailty distributions via kernel density estimation. We assessed the association between frailty category and receipt of surgery via logistic regression.
Results
Among 8237 patients (mean age = 78 years, 57.6% female, 77.4% White race), 2486 (30.2%) underwent curative-intent pancreatectomy. Surgery was performed in 39.7% of nonfrail and 30.7% of prefrail patients; after adjustment, prefrail patients had similar odds of resection compared to nonfrail patients (adjusted odds ratio [OR] = 1.01, 95% confidence interval [CI] = 0.89 to 1.16). In contrast, 15.8% of mildly frail and 5.7% of moderately to severely frail patients underwent surgery; both groups had statistically significant lower odds of surgery vs nonfrail patients (mildly frail adjusted OR = 0.53, 95% CI = 0.43 to 0.66; moderately to severely frail adjusted OR = 0.18, 95% CI = 0.11 to 0.30).
Conclusions
Remarkably, less than 40% of nonfrail patients with nonmetastatic pancreatic ductal adenocarcinoma underwent surgery, suggesting that curative-intent treatment remains infrequent even among lower-risk adults. Frailty is also common and strongly associated with decreased odds of surgical resection.
Introduction
Pancreatic ductal adenocarcinoma was the third leading cause of cancer-related deaths in the United States in 2024.1 Pancreatic ductal adenocarcinoma is common in patients of advanced age, as the median age at diagnosis is 70 years.2 Older adults are a heterogeneous population with varying comorbid conditions, and chronologic age alone is insufficient in explaining differences in treatment selection and outcomes among patients with pancreatic ductal adenocarcinoma. Frailty is a measure that attempts to describe the heterogeneity of aging by evaluating a patient’s physiologic reserve and ability to resist environmental stressors.3 It is a superior predictor of morbidity and mortality compared with chronological age alone.4 With respect to cancer-related outcomes, frailty is associated with increased mortality, chemotherapy intolerance, and treatment-related toxicity across a wide range of malignancies.5-7 In patients with pancreatic ductal adenocarcinoma, the relation between frailty and receipt of curative-intent treatment remains unexplored at the population level. This knowledge gap is noteworthy given the considerable toxicity and risks associated with many pancreatic ductal adenocarcinoma treatment strategies, including major abdominal surgery.
For patients with nonmetastatic pancreatic ductal adenocarcinoma, the current standard of care and optimal treatment strategy is surgery in combination with multiagent systemic chemotherapy.8-10 Advances in surgical techniques and the evolution of chemotherapy administration have broadened the patient population that is offered a potentially curative approach to treatment. However, although some older adults can safely undergo pancreatectomy, increased age is associated with a higher risk of postoperative morbidity and mortality.11,12 Regardless of age, population-based studies have demonstrated that in patients with resectable pancreatic ductal adenocarcinoma, more than one-third of patients will never undergo surgery.13,14 Historically, landmark clinical trials for treatment of pancreatic ductal adenocarcinoma have also excluded frail patients.15-17 Therefore, it remains unclear how frequently patients with higher frailty undergo curative-intent surgical resection in the real world.
The gold standard for frailty assessment involves a comprehensive clinical evaluation, yet the development of validated and broadly accessible metrics to assess patient frailty in claims data has enabled population-level analyses of frailty and frailty-related outcomes. The claims-based frailty index is a widely used measure that implements a deficit accumulation approach to quantify patient vulnerability by generating a score from 0 to 1 (higher score reflects greater frailty) and represents the culmination of numerous health deficits across multiple domains. The claims-based frailty index has outperformed other indices in measuring the extent of patient frailty not explained by age and sex alone.18 The claims-based frailty index has been extensively validated and frequently used in cancer-related research.19-22 Using the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER)–Medicare linked claims database, the objective of this study is to describe the distribution of frailty in the nonmetastatic pancreatic ductal adenocarcinoma population and to evaluate the association between frailty as defined by the claims-based frailty index and receipt of curative-intent pancreatic resection. These data will provide a better understanding of the prevalence and associations between frailty and receipt of curative-intent therapy, providing population-level insight into how physiologic vulnerability shapes treatment patterns in older adults with nonmetastatic pancreatic ductal adenocarcinoma.
Methods
Study design and setting
We performed a quantitative retrospective cohort analysis using SEER-Medicare linked data. The Yale human investigation committee determined that this study did not constitute human participants research.
Study sample
Patients were included if they were Medicare beneficiaries who were diagnosed with pancreatic ductal adenocarcinoma from 2013 to 2019, aged 66 years or older at diagnosis, and enrolled in continuous Medicare Parts A, B, and D fee-for-service coverage from 1 year prior to diagnosis through end-of-life or 1 year after diagnosis, whichever came first. Patients with metastatic disease at diagnosis; missing stage or sociodemographic information; another cancer diagnosis in the 2 years preceding their pancreatic ductal adenocarcinoma diagnosis; a prior history of metastatic cancer; treatment with radiotherapy or chemotherapy in the year prior to pancreatic ductal adenocarcinoma diagnosis; unknown month of pancreatic ductal adenocarcinoma diagnosis; or diagnosis by autopsy or death certificate were excluded (Figure S1).
Study variables
The primary study outcome was receipt of curative-intent surgery in the 12-month period following pancreatic ductal adenocarcinoma diagnosis. Curative-intent surgery was defined as partial, distal, or near total pancreatectomy (Current Procedural Terminology [CPT] codes: 48140-48146), pancreaticoduodenectomy (CPT codes: 48150, 48152, 48153, 48154), or total pancreatectomy (CPT codes: 48155-48160). Frailty was defined using the claims-based frailty index, which utilizes Medicare claims data in the year prior to diagnosis to generate a continuous score using a deficit accumulation frailty index model.23 The claims-based frailty index was developed and validated as a predictor of death, disability, recurrent falls, and health-care utilization using the Medicare Current Beneficiary Survey in 2006. The claims-based frailty index is calculated by assigning predefined weights to International Classification of Diseases diagnosis codes, CPT codes, and Healthcare Common Procedure Coding System procedure codes.
Based on prior studies, we stratified patients into 4 groups based on their prediagnosis claims-based frailty index: nonfrail (claims-based frailty index < 0.15), prefrail (claims-based frailty index ≥ 0.15 and < 0.25), mildly frail (claims-based frailty index ≥ 0.25 and < 0.35), and moderately to severely frail (claims-based frailty index ≥ 0.35).24 Lower claims-based frailty index categories reflect less claims-based evidence of physiologic vulnerability and functional impairment, while higher categories reflect increasing accumulation of health deficits.23,25
Patient characteristics included age, sex, race and ethnicity, area-level median household income quintile, Elixhauser Comorbidity Index, tumor location, T staging, and year of diagnosis. Clinical characteristics for descriptive purposes only included number of lymph nodes evaluated, number of positive lymph nodes, N staging, receipt of chemotherapy, and receipt of radiotherapy. For T staging, the groups were divided into T0-T3 tumors vs T4 tumors as a T4 tumor was considered a surrogate for locally advanced pancreatic ductal adenocarcinoma with extrapancreatic extension and/or vascular involvement.26
Statistical analyses
We used kernel density estimation to assess the distribution of continuous claims-based frailty index scores across the total study population and subpopulations defined by surgical status. We evaluated differences in patient characteristics across the 4 claims-based frailty index categories (nonfrail, prefrail, mildly frail, moderately to severely frail) using χ2 tests for categorical variables and one-way analysis of variance for continuous variables. We assessed treatment pattern differences using descriptive analyses only (χ2 tests) including receipt of chemotherapy (yes or no) and receipt of radiotherapy (yes or no) across the 4 claims-based frailty index categories.
We fit a logistic regression model to examine the unadjusted association between claims-based frailty index category (as independent variable) and surgical resection (as outcome) in patients with pancreatic ductal adenocarcinoma. We then fit a multivariable logistic regression model with robust standard errors to evaluate the relation between the primary independent variable of interest (claims-based frailty index category) and the receipt of surgery, adjusting for clinical and sociodemographic characteristics. Logistic regression model results are presented as odds ratios (ORs) with corresponding 95% confidence intervals (CIs) for the association between claims-based frailty index category and receipt of surgery. Additionally, individual predicted probabilities of surgery were estimated from the multivariable model using observed covariate values and plotted across claims-based frailty index categories.
The regression model was examined for collinearity by assessing variance inflation factors. Covariates with variance inflation factors greater than 3 were considered moderately collinear and evaluated further for model inclusion.27 We also performed several sensitivity analyses to account for assumptions made during the regression analyses. We performed a series of regression analyses based on all possible combinations of the 8 adjustment covariates in the model (ie, 256 combinations). We also calculated the E-values for the adjusted odds ratio estimates and confidence intervals between each of the claims-based frailty index categories and surgical resection, which quantify how strong an unmeasured variable would need to be associated with the claims-based frailty index category and surgery to nullify the claims-based frailty index category–surgery association.28 Last, we fit unadjusted and adjusted logistic regression models for surgery where claims-based frailty index was treated as a continuous variable and report odds ratios corresponding to 1 interquartile range (IQR) change in claims-based frailty index. All analyses were conducted using STATA version 18.0 and R version 4.3.1.
Results
Our study sample included 8237 patients (Figure S1). The median age was 78 years, and most patients were female (57.6%) and White race (77.4%) and had tumors in the head of the pancreas (63.3%) (Table 1). There were 2585 (31.4%) nonfrail patients, 4043 (49.1%) prefrail patients, 1256 (15.2%) mildly frail patients, and 353 (4.3%) moderately to severely frail patients in the total cohort with a median claims-based frailty index score of 0.18.
Table 1.
Sociodemographic and clinical characteristics of patients with nonmetastatic pancreatic ductal adenocarcinoma (n = 8237), by CFI category.a
| Characteristic | Full sample, No. (%)b | CFI frailty strata |
||||
|---|---|---|---|---|---|---|
| Nonfrail CFI 1, No. (%) | Prefrail CFI 2, No. (%) | Mildly frail CFI 3, No. (%) | Moderately to severely frail CFI 4, No. (%) | P | ||
| Total | 8237 | 2585 | 4043 | 1256 | 353 | |
| Age, median (IQR), y | 78 (72-84) | 75 (70-80) | 78 (73-84) | 81 (75-87) | 83 (76-88) | <.001* |
| Gender | <.001* | |||||
| Male | 3489 (42.4) | 1272 (36.4) | 1695 (48.6) | 418 (12.0) | 104 (3.0) | |
| Female | 4748 (57.6) | 1313 (27.7) | 2348 (49.5) | 838 (17.6) | 249 (5.2) | |
| Race and ethnicity | <.001* | |||||
| Asian | 479 (5.8) | 169 (35.3) | 240 (50.1) | 56 (11.7) | 14 (2.9) | |
| Black | 581 (7.1) | 123 (21.2) | 289 (49.7) | 118 (20.3) | 51 (8.8) | |
| Hispanic | 618 (7.5) | 180 (29.1) | 297 (48.1) | 108 (17.5) | 33 (5.3) | |
| White | 6376 (77.4) | <2042 (<32.0) | 3139 (49.2) | 951 (14.9) | >244 (>3.8) | |
| Unknown | 183 (2.2) | >71 (>38.8) | 78 (42.6) | 23 (12.6) | <11 (<6.0) | |
| Area-level median household income | <.001* | |||||
| Lowest quintile | 1630 (19.8) | 397 (24.3) | 829 (50.9) | 305 (18.7) | 99 (6.1) | |
| Second quintile | 1663 (20.2) | 458 (27.5) | 860 (51.7) | 271 (16.3) | 74 (4.5) | |
| Middle quintile | 1587 (19.3) | 520 (32.8) | 749 (47.2) | 243 (15.3) | 75 (4.7) | |
| Fourth quintile | 1668 (20.2) | 566 (33.9) | 810 (48.6) | 242 (14.5) | 50 (3.0) | |
| Highest quintile | 1689 (20.5) | 644 (38.1) | 795 (47.1) | 195 (11.5) | 55 (3.3) | |
| Elixhauser Comorbidity Index | <.001* | |||||
| 0 conditions | 2150 (26.1) | >1482 (>68.9) | 617 (28.7) | 40 (1.9) | <11 (<0.5) | |
| 1-2 conditions | 3324 (40.4) | <972 (<29.2) | 2046 (61.5) | 288 (8.7) | >18 (>0.5) | |
| ≥3 conditions | 2763 (33.5) | 131 (4.7) | 1380 (50.0) | 928 (33.6) | 324 (11.7) | |
| Tumor location | ||||||
| No surgery | <.001* | |||||
| Head | 3497 (60.8) | 984 (28.2) | 1704 (48.7) | 620 (17.7) | 189 (5.4) | |
| Body, tail | 878 (15.3) | 276 (31.4) | 439 (50.0) | 126 (14.4) | 37 (4.2) | |
| Other | 1376 (23.9) | <302 (21.9) | 660 (48.0) | 311 (22.6) | >103 (>7.5) | |
| Surgery | .033* | |||||
| Head | 1720 (69.2) | 733 (42.6) | 853 (49.6) | 121 (7.0) | 13 (0.8) | |
| Body, tail | 510 (20.5) | >192 (>37.6) | 253 (49.6) | 54 (10.6) | <11 (<2.2) | |
| Other | 256 (10.3) | 98 (38.3) | 134 (52.3) | 24 (9.4) | 0 (0) | |
Abbreviation: CFI = claims-based frailty index.
χ2 tests were used for categorical variables, and 1-way analysis of variance was used for age (continuous variable).
Percentages in the CFI frailty strata represent row percentages. Percentages in the full sample column represent the proportion of the total cohort.
P < .05 is statistically significant.
The distribution of claims-based frailty index scores differed between the surgical and nonsurgical populations (Figure 1). Surgical patients tended toward lower claims-based frailty index scores with a median of 0.16. In contrast, nonsurgical patients had a median claims-based frailty index of 0.19 and a heavier right tail toward higher frailty. The surgical patients also had a statistically significant lower mean claims-based frailty index score than the nonsurgical patients (0.17 vs 0.20, respectively; P < .001).
Figure 1.
Distribution of frailty using the claims-based frailty index, by surgery status, in nonmetastatic pancreatic ductal adenocarcinoma. This figure shows the kernel density estimate of the distribution of the claims-based frailty index among individuals who underwent surgery (n = 2486), individuals who did not undergo surgery (n = 5751), and the total cohort (n = 8237).
There were statistically significant differences in patients’ sociodemographic characteristics across claims-based frailty index categories, including age, sex, race and ethnicity, median income quintile, Elixhauser Comorbidity Index, and tumor location (Table 1). Older patients, females, patients of Black race, patients residing in a lower median income region, and those with more medical comorbidities were more frequently moderately to severely frail.
Treatment patterns stratified by frailty
Curative-intent surgical resection was performed in 2486 (30.2%) patients. The nonfrail and prefrail categories had a higher proportion of patients undergoing curative surgery than those in the mildly frail or moderately to severely frail categories (Figure 2A). A similar pattern was observed for receipt of chemotherapy or radiotherapy, with increasing frailty resulting in a lower percentage of patients receiving either treatment. Regarding combined treatments, most patients (79.5%) who underwent curative-intent surgery received at least 1 cycle of chemotherapy, regardless of frailty (Figure 2B). However, receipt of chemotherapy in addition to surgery differed significantly across frailty categories, with lower percentages observed among increasingly frail groups (P < .001).
Figure 2.
Treatment characteristics of nonmetastatic pancreatic ductal adenocarcinoma patients, by claims-based frailty index category. A) Demonstrates the percentage of patients within each frailty stratum who underwent curative surgery, chemotherapy, or radiotherapy. B) Stacked bar graph that represents all patients within each frailty stratum who underwent curative surgery, stratified by receipt of chemotherapy.
Tumor characteristics did not vary across the 4 frailty categories in those who underwent curative-intent pancreatic resection. Most patients had a T3 tumor, were N0 or N1, and had more than 10 lymph nodes examined at the time of surgery (Table S1). Across all frailty categories, 18.2% of patients had T4 tumors, also described as locally advanced tumors, indicating involvement of major adjacent vascular structures. The prevalence of T4 tumors did not vary across frailty groups, but the proportion undergoing surgery differed. Of the T4 patients, 11.6% underwent curative-intent surgery, and 95.6% of these surgical patients were either nonfrail or prefrail.
Association between frailty and receipt of surgery
Higher claims-based frailty index category was associated with a statistically significant decrease in the odds of receiving curative-intent pancreatic resection compared to nonfrail patients (prefrail: OR = 0.67, 95% CI = 0.61 to 0.74; mildly frail: OR = 0.29, 95% CI = 0.24 to 0.34; moderately to severely frail: OR = 0.09, 95% CI = 0.06 to 0.14) (Table 2).
Table 2.
Logistic regression models evaluating the relation between frailty and receipt of surgery.
| Bivariate |
Multivariable |
|||
|---|---|---|---|---|
| Variable | OR (95% CI) | P | aOR (95% CI) | P |
| Frailty category | ||||
| Nonfrail | 1.00 (Referent) | — | 1.00 (Referent) | — |
| Prefrail | 0.67 (0.61 to 0.74) | <.001* | 1.01 (0.89 to 1.16) | .833 |
| Mildly frail | 0.29 (0.24 to 0.34) | <.001* | 0.53 (0.43 to 0.66) | .001* |
| Moderately to severely frail | 0.09 (0.06 to 0.14) | <.001* | 0.18 (0.11 to 0.30) | <.001* |
| Continuous CFIa | 0.50 (0.46 to 0.53) | <.001* | 0.62 (0.56 to 0.68) | <.001* |
Abbreviations: aOR = adjusted odds ratio; CFI = claims-based frailty index robust standard errors used in the multivariable model; CI = confidence interval; OR = odds ratio.
Odds ratios for continuous variables are expressed per 1 interquartile range increase in the CFI score.
P < .05 is statistically significant.
After adjusting for clinical and sociodemographic characteristics, patients in the prefrail category had similar odds of undergoing surgical resection compared with nonfrail patients (adjusted OR = 1.01, 95% CI = 0.89 to 1.16) (Table 2; see also Figure S2). Patients in the mildly frail and moderately to severely frail categories, however, had statistically significant decreased odds compared with those in the nonfrail category of undergoing surgical resection (mildly frail: adjusted OR = 0.53, 95% CI = 0.43 to 0.66; moderately to severely frail: adjusted OR = 0.18, 95% CI = 0.11 to 0.30). The median of the estimated probabilities of surgery was 0.44 (IQR = 0.20-0.57) among nonfrail patients, 0.29 (IQR = 0.11-0.48) in prefrail, 0.12 (IQR = 0.04-0.26) in mildly frail, and 0.03 (IQR = 0.01-0.10) in moderately to severely frail (Figure 3). Covariates in the adjusted model that were associated with a decreased odds of surgery included increased age, Black race, having at least 3 comorbid conditions, and T4 stage (Table S2). There was no statistically significant interaction between age and frailty when evaluating their association with receipt of surgery. Alternatively, covariates associated with an increased odds of surgery included a higher area-level median household income and a body and/or tail pancreatic tumor.
Figure 3.
Distribution of estimated probabilities of surgery by frailty category. The box plots show the estimated probabilities of surgery among individuals in each frailty group: nonfrail (n = 2585), prefrail (n = 4043), mildly frail (n = 1256), and moderately to severely frail (n = 353). The boxes represent the first and third quartiles with a horizontal line at the median. The whiskers extend to the largest value within 1.5 times the interquartile range from the box. Values beyond the whiskers are shown as individual points. The estimated probabilities were obtained from the multivariable logistic regression model.
All variance inflation factors in the adjusted multivariable logistic regression model were less than 3, indicating no substantial multicollinearity. When removing the Elixhauser Comorbidity Index as a covariate, the model again demonstrated a significantly lower odds of surgery for mildly frail (adjusted OR = 0.44, 95% CI = 0.36 to 0.53) and moderately to severely frail (adjusted OR = 0.15, 95% CI = 0.09 to 0.23) patients compared with nonfrail patients with no statistically significant association with receipt of surgery for prefrail patients (adjusted OR = 0.91, 95% CI = 0.81 to 1.02).
The results of the sensitivity analyses including each possible choice of adjustment covariates in the model are shown in Figure S3. The adjusted odds ratios and corresponding 95% confidence intervals for the mildly frail and moderately to severely frail categories were always below 1.0, whereas the adjusted odds ratios and 95% confidence intervals for the prefrail category included 1.0 for certain combinations of covariates. The E-values for the adjusted odds ratio estimates in the adjusted model were 1.09 for the prefrail, 2.08 for the mildly frail, and 4.09 for the moderately to severely frail.
When treating claims-based frailty index as a continuous variable in the logistic regression analyses, the adjusted odds ratio for surgery was 0.62 (95% CI = 0.56 to 0.68) for 1 IQR increase in the claims-based frailty index, that is, a change from 0.14 (25th percentile) to 0.24 (75th percentile) in the claims-based frailty index.
Discussion
In our study of 8237 patients with nonmetastatic pancreatic ductal adenocarcinoma, 68.6% of patients were found to have some degree of frailty, while only 31.4% of patients were classified as nonfrail. The proportion of frail patients in this nonmetastatic pancreatic ductal adenocarcinoma population is much higher than that of the general Medicare population (52.7% nonfrail, 47.3% prefrail or higher).25 At the population level, frailty had a substantial negative association with receipt of surgical resection. However, even when limited to nonfrail patients, only 40% underwent surgery. Moreover, in those who did undergo surgical resection, fewer patients received systemic chemotherapy as frailty increased.
Frailty is an aging-related syndrome that can exert a profound effect on each stage of the care continuum for patients with cancer. While frailty metrics are beginning to emerge in clinical research, many clinical trials still utilize performance status alone as a surrogate for health status. The landmark clinical trials for pancreatic ductal adenocarcinoma that established the current standard of care for optimal treatment have required patients to have an Eastern Cooperative Oncology Group Performance Status of 0-1 (fully active to restricted in strenuous activity only) or Karnofsky Performance Status of at least 50% (functionally capable and able to care for most personal needs) to be eligible for enrollment.15-17,29-31
It is recognized that clinical trial data do not adequately represent the general population, nor are the results generalizable to the real world.32 In this study of a representative and real-world population of Medicare recipients in the United States, only 31.4% of patients were classified as nonfrail and therefore potentially able to qualify for clinical trial enrollment. The largest subpopulation identified in this study was prefrail patients. These patients have claims suggesting emerging deficits, such as a history of falls, mild cognitive decline, or early dependency. When combined with the frail population, these data suggest that as many as two-thirds of nonmetastatic pancreatic ductal adenocarcinoma patients are not represented by the typical clinical trial population. Thus, it is difficult to decipher the optimal modality of treatment for these patients as well as the expected outcomes of these treatments.
Advances in surgical technique and preoperative patient optimization have enabled curative-intent surgical resection to be offered more widely. It is striking, therefore, that only 30.2% of patients underwent curative-intent surgery in this nonmetastatic population. Even when limited to nonfrail patients with tumors lacking extrapancreatic extension or arterial invasion (T0-T3), only 45.2% underwent surgical resection. Although this subpopulation does not rule out venous involvement, this group most closely approximates patients considered “resectable” or “borderline resectable” by the current National Comprehensive Cancer Network guidelines with the most favorable survival following surgery and chemotherapy. Notably, nearly 20% of surgical patients in this group did not receive a single cycle of chemotherapy. These data suggest that adherence to treatment guidelines for optimal outcomes in patients with nonmetastatic pancreatic ductal adenocarcinoma is imperfect at best.
We found that, after adjusting for sociodemographic and clinical characteristics, prefrail patients had similar odds of surgical resection compared with nonfrail patients, while mildly or moderately to severely frail patients continued to have substantially decreased odds of surgical resection. Several studies have evaluated the effect of chronological age on postoperative outcomes and found that advanced age is associated with a statistically significant higher risk of postoperative morbidity and mortality compared with younger patients but that surgery could safely be accomplished in select patients.11,12,33 This population is often the target for prehabilitation programs prior to surgery to improve physical function and functional capacity.34,35 In aggregate, these data suggest that surgeons were amenable to offering surgery to prefrail patients if, after comprehensive evaluation, the risk of operating could be favorably modified or addressed.
A major novel finding of this study is that regardless of frailty, only a small proportion of patients received curative-intent treatment for nonmetastatic pancreatic ductal adenocarcinoma. Existing studies have highlighted the low incidence of surgery in patients with pancreatic ductal adenocarcinoma, attributing this finding to advanced disease stage and patient characteristics such as age or frailty.14,36,37 In this study, we conclusively demonstrate that even in candidates seemingly best suited for curative-intent resection (nonmetastatic and nonfrail), surgery is still surprisingly infrequent. Further research is needed to understand the reason for this finding, which is likely multifactorial.
In addition to patient- and tumor-level characteristics, treatment selection is likely influenced by unmeasured provider- and system-level determinants. For instance, in this study, we found that patients residing in a region with a higher median income had higher odds of undergoing surgical resection, demonstrating the interplay of underlying socioeconomic factors with patient care. Prior work has also demonstrated substantial geographic- and hospital-level variation in patient access to curative-intent resection, suggesting that referral patterns, institutional practice variations, and multidisciplinary treatment pathways may contribute to the observed differences in treatment utilization.38 At the individual level, decisions regarding curative-intent surgery are complex and may be influenced by surgeon experience, institutional support for high-risk operative management, rehabilitation resources, caregiver support, and patient preferences. Although unmeasured confounding remains possible, sensitivity analyses using E-values suggested that progressively stronger unmeasured confounding would be required to fully explain the observed associations between higher frailty categories and lower odds of surgical resection (Figure S3).28 Moreover, these findings were not limited to surgery because receipt of systemic therapy and radiation were also much lower than expected and decreased across increasing frailty categories.
There are several limitations of this study. This analysis was performed on observational, retrospective data and is therefore limited by selection bias and retrospectively defined parameters. For example, SEER-Medicare does not capture data such as radiographic findings, medical decision-making, and intent of treatment. Additionally, although SEER-Medicare covers a large portion of the US population, it does not represent all regions or demographic groups. However, claims-based datasets such as SEER-Medicare more accurately capture treatment data compared with other widely used oncology datasets such as the National Cancer Database and are superior at capturing comorbid conditions that constitute the basis of the frailty index.39,40
In summary, frailty is a statistically significant factor associated with lower odds of curative-intent surgical resection in patients with nonmetastatic pancreatic ductal adenocarcinoma. With more than two-thirds of the nonmetastatic pancreatic ductal adenocarcinoma population falling into at least a prefrail category, it is imperative to understand how frailty is associated with treatment patterns and outcomes. There remains a substantial void in the literature around frailty and its association with cancer-directed treatment and clinical outcomes in patients with localized pancreatic ductal adenocarcinoma. Future work will focus on addressing this knowledge gap by analyzing patterns of care, clinical outcomes of specific treatment choices, and overall survival in this patient population.
Supplementary Material
Acknowledgments
The collection of the California cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute (NCI) Surveillance, Epidemiology and End Results Program under contract N01-PC-35136 awarded to the Northern California Cancer Center, contract N01-PC-35139 awarded to the University of Southern California, and contract N02-PC-15105 awarded to the Public Health Institute; and the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement #U55/CCR921930-02 awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and endorsement by the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should be inferred. The authors acknowledge the efforts of the Applied Research Program, NCI; the Office of Research, Development and Information, Centers for Medicare and Medicaid Services; Information Management Services, Inc.; and the SEER Program tumor registries in the creation of the SEER-Medicare database. The interpretation and reporting of the SEER-Medicare data are the sole responsibility of the authors.
The funder had no role in the design of the study; the collection, analysis, or interpretation of the data; or the writing of the manuscript and decision to submit it for publication.
Contributor Information
Lauren Raymond-King, Department of Surgery, Division of Surgical Oncology, Yale School of Medicine, New Haven, CT, United States; National Clinician Scholars Program (NCSP), Yale University, New Haven, CT, United States.
Sean McGrath, Department of Biostatistics, Yale School of Public Health, New Haven, CT, United States.
Bhramar Mukherjee, Department of Biostatistics, Yale School of Public Health, New Haven, CT, United States.
Yiran Wang, Department of Biostatistics, Yale School of Public Health, New Haven, CT, United States.
Harsh Parikh, Department of Biostatistics, Yale School of Public Health, New Haven, CT, United States.
John Rothen, Cancer Outcomes, Public Policy, and Effectiveness Research (COPPER) Center, Yale Cancer Center and Yale University School of Medicine, New Haven, CT, United States.
Pamela R Soulos, Cancer Outcomes, Public Policy, and Effectiveness Research (COPPER) Center, Yale Cancer Center and Yale University School of Medicine, New Haven, CT, United States.
Cary P Gross, National Clinician Scholars Program (NCSP), Yale University, New Haven, CT, United States; Cancer Outcomes, Public Policy, and Effectiveness Research (COPPER) Center, Yale Cancer Center and Yale University School of Medicine, New Haven, CT, United States; Department of Internal Medicine, Yale School of Medicine, New Haven, CT, United States.
John W Kunstman, Department of Surgery, Division of Surgical Oncology, Yale School of Medicine, New Haven, CT, United States.
Author contributions
Lauren Raymond-King (Conceptualization, Formal analysis, Methodology, Writing—original draft), Sean McGrath (Conceptualization, Formal analysis, Methodology, Visualization, Writing—review & editing), Bhramar Mukherjee (Conceptualization, Methodology, Writing—review & editing), Yiran Wang (Methodology, Writing—original draft), Harsh Parikh (Methodology, Writing—review & editing), John Rothen (Data curation, Resources, Software, Writing—review & editing), Pamela R. Soulos (Conceptualization, Data curation, Methodology, Resources, Software, Writing—review & editing), Cary P. Gross (Conceptualization, Methodology, Writing—review & editing), and John W. Kunstman (Conceptualization, Methodology, Supervision, Writing—review & editing)
Supplementary material
Supplementary material is available at JNCI Cancer Spectrum online.
Funding
Lauren Raymond-King is supported by the National Institute of Aging and the National Institute of Health (T32AG019134). Cary Gross has received research funding, through Yale, from the National Comprehensive Cancer Network (NCCN) Oncology Research Program (with funding provided to the NCCN by AstraZeneca Pharmaceuticals LP), as well as funding from Johnson and Johnson to help devise and implement new approaches to sharing clinical trial data. Sean McGrath, Bhramar Mukherjee, Yiran Wang, Harsh Parikh, John Rothen, Pamela Soulos, and John Kunstman have nothing to disclose.
Conflicts of interest
The authors have no relevant conflicts of interest to disclose.
Data availability
The data underlying this article were provided by SEER-Medicare by permission via a data use agreement. Data will be shared on request to the corresponding author with permission of SEER-Medicare.
References
- 1. Surveillance E, and End Results Program. Cancer Stat Facts: Common Cancer Sites. National Cancer Institute; 2024.
- 2. Surveillance E, and End Results Program. Cancer Stat Facts: Pancreatic Cancer. National Cancer Institute; 2024.
- 3. Song X, Mitnitski A, Rockwood K. Prevalence and 10-year outcomes of frailty in older adults in relation to deficit accumulation. J Am Geriatr Soc. 2010;58:681-687. 10.1111/j.1532-5415.2010.02764.x [DOI] [PubMed] [Google Scholar]
- 4. Gu C, Lu A, Lei C, et al. Frailty index is useful for predicting postoperative morbidity in older patients undergoing gastrointestinal surgery: a prospective cohort study. BMC Surg. 2022;22:57. 10.1186/s12893-022-01471-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Kristjansson SR, Rønning B, Hurria A, et al. A comparison of two pre-operative frailty measures in older surgical cancer patients. J Geriatr Oncol. 2012;3:1-7. 10.1016/j.jgo.2011.09.002 [DOI] [Google Scholar]
- 6. Clough-Gorr KM, Thwin SS, Stuck AE, et al. Examining five- and ten-year survival in older women with breast cancer using cancer-specific geriatric assessment. Eur J Cancer. 2012;48:805-812. 10.1016/j.ejca.2011.06.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Chappidi MR, Kates M, Patel HD, et al. Frailty as a marker of adverse outcomes in patients with bladder cancer undergoing radical cystectomy. Urol Oncol 2016;34:256.e1-256-e6. 10.1016/j.urolonc.2015.12.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Neoptolemos JP, Stocken DD, Friess H, et al. ; European Study Group for Pancreatic Cancer. A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer. N Engl J Med. 2004;350:1200-1210. 10.1056/NEJMoa032295 [DOI] [PubMed] [Google Scholar]
- 9. Neoptolemos JP, Cox TF. Bayesian analysis unravels pancreas-cancer adjuvant therapy. Lancet Oncol. 2013;14:1034-1035. 10.1016/S1470-2045(13)70403-0 [DOI] [PubMed] [Google Scholar]
- 10. Tempero MA, Malafa MP, Al-Hawary M, et al. Pancreatic adenocarcinoma, version 2.2021, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2021;19:439-457. 2021 10.6004/jnccn.2021.0017 [DOI] [PubMed] [Google Scholar]
- 11. Haigh PI, Bilimoria KY, DiFronzo LA. Early postoperative outcomes after pancreaticoduodenectomy in the elderly. Arch Surg. 2011;146:715-723. 10.1001/archsurg.2011.115 [DOI] [PubMed] [Google Scholar]
- 12. Finlayson E, Fan Z, Birkmeyer JD. Outcomes in octogenarians undergoing high-risk cancer operation: a national study. J Am Coll Surg. 2007;205:729-734. 10.1016/j.jamcollsurg.2007.06.307 [DOI] [PubMed] [Google Scholar]
- 13. Dimou F, Sineshaw H, Parmar AD, et al. Trends in receipt and timing of multimodality therapy in early-stage pancreatic cancer. J Gastrointest Surg. 2016;20:93-103; discussion 103. 10.1007/s11605-015-2952-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Heller DR, Nicolson NG, Ahuja N, et al. Association of treatment inequity and ancestry with pancreatic ductal adenocarcinoma survival. JAMA Surg. 2020;155:e195047. 10.1001/jamasurg.2019.5047 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. Jama. 2007;297:267-277. 10.1001/jama.297.3.267 [DOI] [PubMed] [Google Scholar]
- 16. Neoptolemos JP, Palmer DH, Ghaneh P, et al. ; European Study Group for Pancreatic Cancer. Comparison of adjuvant gemcitabine and capecitabine with gemcitabine monotherapy in patients with resected pancreatic cancer (ESPAC-4): a multicentre, open-label, randomised, phase 3 trial. Lancet. 2017;389:1011-1024. 10.1016/S0140-6736(16)32409-6 [DOI] [PubMed] [Google Scholar]
- 17. Sohal DPS, Duong M, Ahmad SA, et al. Efficacy of perioperative chemotherapy for resectable pancreatic adenocarcinoma: a phase 2 randomized clinical trial. JAMA Oncol. 2021;7:421-427. 10.1001/jamaoncol.2020.7328 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Kim DH, Patorno E, Pawar A, et al. Measuring frailty in administrative claims data: comparative performance of four claims-based frailty measures in the U.S. Medicare Data. J Gerontol A Biol Sci Med Sci. 2020;75:1120-1125. 10.1093/gerona/glz224 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Minami CA, Jin G, Freedman RA, et al. Association of surgery with frailty status in older women with early-stage breast cancer. JAMA Surg. 2023;158:664-666. 10.1001/jamasurg.2022.8146 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Munir MM, Woldesenbet S, Pawlik TM. Trajectory analysis of healthcare use before and after gastrointestinal cancer surgery. J Am Coll Surg. 2025;240:24-33. 10.1097/XCS.0000000000001212 [DOI] [PubMed] [Google Scholar]
- 21. Gohil S, Chen H, Lewis WE, et al. Factors associated with the use of immune checkpoint inhibitors in older adults with metastatic non-small cell lung cancer and pre-existing autoimmune disease: a SEER-Medicare study. J Geriatr Oncol. 2024;15:102071. 10.1016/j.jgo.2024.102071 [DOI] [PubMed] [Google Scholar]
- 22. Wong ML, Lichtman SM, Morrow GR, et al. Geriatric oncology health services research: cancer and Aging Research Group infrastructure core. J Geriatr Oncol. 2020;11:350-354. 10.1016/j.jgo.2019.07.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Kim DH, Schneeweiss S, Glynn RJ, et al. Measuring frailty in Medicare data: development and validation of a claims-based frailty index. J Gerontol A Biol Sci Med Sci. 2018;73:980-987. 10.1093/gerona/glx229 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Lee KC, Streid J, Sturgeon D, et al. The impact of frailty on long-term patient-oriented outcomes after emergency general surgery: a retrospective cohort study. J Am Geriatr Soc. 2020;68:1037-1043. 10.1111/jgs.16334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Kim DH, Glynn RJ, Avorn J, et al. Validation of a claims-based frailty index against physical performance and adverse health outcomes in the health and retirement study. J Gerontol A Biol Sci Med Sci. 2019;74:1271-1276. 10.1093/gerona/gly197 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Shin DW, Kim J. The American Joint Committee on Cancer 8th edition staging system for the pancreatic ductal adenocarcinoma: is it better than the 7th edition? Hepatobiliary Surg Nutr. 2020;9:98-100. 10.21037/hbsn.2019.08.06 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Vatcheva KP, Lee M, McCormick JB, et al. Multicollinearity in regression analyses conducted in epidemiologic studies. Epidemiology (Sunnyvale). 2016;6:227. 10.4172/2161-1165.1000227 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-Value. Ann Intern Med. 2017;167:268-274. 10.7326/M16-2607 [DOI] [PubMed] [Google Scholar]
- 29. 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. 10.1001/jama.2013.279201 [DOI] [PubMed] [Google Scholar]
- 30. Azam F, Latif MF, Farooq A, et al. Performance status assessment by using ECOG (Eastern Cooperative Oncology Group) score for cancer patients by oncology healthcare professionals. Case Rep Oncol. 2019;12:728-736. 10.1159/000503095 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Péus D, Newcomb N, Hofer S. Appraisal of the Karnofsky Performance Status and proposal of a simple algorithmic system for its evaluation. BMC Med Inform Decis Mak. 2013;13:72. 10.1186/1472-6947-13-72 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Malmivaara A. Generalizability of findings from randomized controlled trials is limited in the leading general medical journals. J Clin Epidemiol. 2019;107:36-41. 10.1016/j.jclinepi.2018.11.014 [DOI] [PubMed] [Google Scholar]
- 33. Shaib WL, Jones JS, Goodman M, et al. Evaluation of treatment patterns and survival outcomes in elderly pancreatic cancer patients: a surveillance, epidemiology, and end results-medicare analysis. Oncologist. 2018;23:704-711. 10.1634/theoncologist.2017-0487 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Ngo-Huang A, Parker NH, Bruera E, et al. Home-based exercise prehabilitation during preoperative treatment for pancreatic cancer is associated with improvement in physical function and quality of life. Integr Cancer Ther. 2019;18:1534735419894061. 10.1177/1534735419894061 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Nepal B, Gupta R, Yadav TD, et al. Semisupervised home-based prehabilitation in patients undergoing pancreaticoduodenectomy: impact on postoperative outcomes. Surgery. 2025;184:109455. 10.1016/j.surg.2025.109455 [DOI] [PubMed] [Google Scholar]
- 36. Riall TS, Sheffield KM, Kuo YF, et al. Resection benefits older adults with locoregional pancreatic cancer despite greater short-term morbidity and mortality. J Am Geriatr Soc. 2011;59:647-654. 10.1111/j.1532-5415.2011.03353.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Amin S, Lucas AL, Frucht H. Evidence for treatment and survival disparities by age in pancreatic adenocarcinoma: a population-based analysis. Pancreas. 2013;42:249-253. 10.1097/MPA.0b013e31825f3af4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Liem MNE, Teeken MD, Ruan M, et al. Geospatial variation in the treatment of Medicare beneficiary pancreatic cancer patients: A SEER-Medicare cohort study. The American Journal of Surgery. 2026;251:116684., 10.1016/j.amjsurg.2025.116684 [DOI] [PubMed] [Google Scholar]
- 39. Noone AM, Lund JL, Mariotto A, et al. Comparison of SEER treatment data with medicare claims. Med Care. 2016;54:e55-64-e64. 10.1097/MLR.0000000000000073 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Wong ML, McMurry TL, Schumacher JR, et al. Comorbidity assessment in the national cancer database for patients with surgically resected breast, colorectal, or lung cancer (AFT-01, -02, -03). J Oncol Pract. 2018;14:e631-e643. 10.1200/JOP.18.00175 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article were provided by SEER-Medicare by permission via a data use agreement. Data will be shared on request to the corresponding author with permission of SEER-Medicare.