Abstract
Background
Pancreatic ductal adenocarcinoma is a clinically challenging malignancy largely because of its chemoresistance. Bacteria within the pancreatic ductal adenocarcinoma microbiome may mediate chemoresistance, suggesting that alteration of the microbiome with antibiotics could improve chemotherapy response.
Methods
We utilized the Surveillance, Epidemiology, and End Results Program–Medicare database to select patients with resected, early-stage pancreatic ductal adenocarcinoma diagnosed between 2007 and 2017. The primary outcome of this study was overall survival. Receipt of antibiotic treatment within 1 month after adjuvant chemotherapy initiation was determined from Medicare claims data. Propensity scores were used to match patients who received antibiotics with patients who did not receive antibiotics. The Kaplan-Meier method was used to calculate 5-year overall survival rates, and Cox regression analysis was used to assess the association between receiving antibiotics and overall survival. All hypotheses were 2 sided.
Results
Of the 712 patients with resected, early-stage pancreatic ductal adenocarcinoma, 629 (88.3%) were treated with adjuvant gemcitabine and 177 (24.9%) received antibiotics in the 1 month following chemotherapy initiation. The mean (SD) age at diagnosis was 73.7 (5.1) years, and patients were mostly women, White, and from metropolitan areas in the northeastern or western United States. A total of 143 propensity score–matched pairs were evaluated. Among patients treated with gemcitabine, antibiotic treatment was associated with a 37% improvement in overall survival and a 30% improvement in cancer-specific survival.
Conclusions
Antibiotic treatment in the 1 month following adjuvant gemcitabine initiation was associated with improved survival. These findings provide additional support for the hypothesis that antibiotic treatment may alter the pancreatic microbiome in a manner that reduces chemoresistance, potentially improving pancreatic ductal adenocarcinoma outcomes.
Introduction
Pancreatic ductal adenocarcinoma is a clinically challenging malignancy with an incidence rate rising at 1% per year.1,2 If it is detected in its early stages, clinical guidelines for localized pancreatic ductal adenocarcinoma have traditionally recommended surgical resection and adjuvant chemotherapy with either gemcitabine-based or fluorouracil-based treatment.3-5 However, the effectiveness of these chemotherapies remains limited, as reflected in the poor 5-year survival rates of 13% for all stages and less than 50% for localized disease.1,6
Pancreatic ductal adenocarcinoma is associated with substantial changes in the pancreatic microbiome, rendering it microbially distinct from that of a noncancerous pancreas.7-10 Multiple preclinical studies have shown that bacteria within the tumor microenvironment can drive chemoresistance through mechanisms, including metabolization11-15 and immune response modulation.16 Specifically, Gammaproteobacteria found in the pancreatic ductal adenocarcinoma microbiome can metabolize and inactivate gemcitabine, suggesting a potential link to chemoresistance.13 Research also indicates that antibiotic-induced bacterial ablation in pancreatic ductal adenocarcinoma mouse models can reshape the tumor microenvironment, improve immune surveillance, and heighten sensitivity to immunotherapy.7 Alteration of the pancreatic microbiome through antibiotic treatment may offer a potential approach to addressing chemoresistance.
We sought to build on small retrospective studies that showcased improved treatment outcomes with antibiotic exposure in pancreatic ductal adenocarcinoma patients.17-20 Our group has previously demonstrated a positive association between receipt of perichemotherapy antibiotics and survival in patients with metastatic pancreatic ductal adenocarcinoma treated with gemcitabine.19 In this study, we investigate the impact of antibiotic use on overall survival in patients with resected, early-stage pancreatic ductal adenocarcinoma who are receiving adjuvant chemotherapy.
Methods
Data source
We used the Surveillance, Epidemiology, and End Results (SEER) Program–Medicare–linked database, encompassing demographic and cancer data from approximately 28% of the US population.21 The SEER database links to Medicare claims data, which details health-care utilization such as hospital stays, prescription medication, and medical procedures. This study was approved by the institutional review board of the Icahn School of Medicine at Mount Sinai with a waiver of informed consent. This decision was based on the recognition of the SEER-Medicare linked database as a limited dataset under Health Insurance Portability and Accountability Act regulations and the determination that the study posed minimal risk. Reporting of this study followed the guidelines outlined in the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline.
Study population
Patients diagnosed with pancreatic ductal adenocarcinoma between January 1, 2007, and December 31, 2017, were identified within the SEER Medicare database. We used SEER data between 2007 and 2017 and Medicare data between 2006 and 2019 to ensure complete capture of Medicare claims from the year before diagnosis to the end of Medicare follow-up. All patients’ pancreatic ductal adenocarcinoma diagnoses were histologically confirmed (Table S1),22-24 and determined to be early stage and eligible for resection based on the American Joint Committee on Cancer (AJCC) Cancer Staging Manual (6th edition). The 6th edition had the most complete data in our cohort and has been validated across multiple studies.25,26 All patients underwent surgery, followed by first-line gemcitabine-based or fluorouracil-based chemotherapy (Figure 1). First-line treatment was defined as the first administration of a chemotherapeutic regimen following diagnosis of pancreatic ductal adenocarcinoma and, following surgical resection, adjuvant gemcitabine or adjuvant fluorouracil-based chemotherapy. We required that patients had received either gemcitabine-based or fluorouracil-based chemotherapy but not both. The majority of patients received combination chemotherapy; few patients had received a single-agent therapeutic. Our study included patients aged 66 years and older to ensure continuous Medicare parts A and B coverage and nonenrollment in a health-care maintenance organization before diagnosis. Part D prescription drug plan enrollment was required during the month following chemotherapy initiation. Patients with more than 1 primary cancer diagnosis, a cancer diagnosis at time of death or on autopsy, or who had died before the end of the antibiotic exposure period were excluded.
Figure 1.
Full and propensity score–matched cohort. Abbreviation: SEER = Surveillance, Epidemiology, and End Results.
Treatment characteristics
Information about the administration of chemotherapy and radiation therapy was obtained from Medicare claims data between 2006 and 2019 (Table S1). To ensure correct classification of adjuvant chemotherapy, patients who initiated chemotherapy before surgery or more than 24 weeks after surgery were excluded (Figure 1). Chemotherapy is typically recommended within 12 weeks of surgery,27,28 though survival is similar if given within 24 weeks of surgery.29-31
Antibiotic exposure
We hypothesized that bacteria within the pancreatic ductal adenocarcinoma microenvironment may confer resistance to chemotherapy. Therefore, antibiotic exposure was defined as an antibiotic prescription for 5 or more days or 1 injectable antibiotic dose within 30 days of chemotherapy initiation. Prior studies have used a similar treatment length19,32 and exposure window.33-35 Patients were classified as having received antibiotics during this time or not having received antibiotics. Only systemic, nonophthalmic, nontopical antibiotic prescriptions and injectable antibiotic prescriptions counted toward antibiotic treatment (Table S2).
Outcomes
The primary focus of this study was overall survival, defined as the time between pancreatic ductal adenocarcinoma diagnosis and death. Patients alive at the end of Medicare follow-up (December 31, 2019) were censored. Cancer-specific survival, defined as the time between pancreatic ductal adenocarcinoma diagnosis and cancer-related death, was a secondary outcome. Patients still alive at the end of SEER registry follow-up (December 31, 2018) or whose cause of death was not cancer related were censored. Other outcomes included survival stratified by antibiotic coverage, class, and route of administration among patients who received antibiotic treatment and gemcitabine.
Statistical analysis
Statistical analysis was conducted between August 1, 2023, and December 15, 2024. Baseline demographic and clinical characteristics were compared between patients who did and did not receive antibiotics during the 1 month following chemotherapy. Standardized mean differences were used to measure covariate distribution between groups.
A multivariable logistic regression model was used to generate propensity scores and model probability of receiving antibiotics based on predetermined demographic and clinical covariates. Covariate information gathered from Medicare claims data included infection during the time of antibiotic receipt (Table S3) and endoscopic retrograde cholangiopancreatography (Table S4) procedures that occurred between diagnosis and the end of the antibiotic exposure window. Endoscopic retrograde cholangiopancreatography can be performed to treat infection, such as cholangitis, but it can also introduce infection into a previously sterile biliary system when used to relieve obstruction. Therefore, it was included as a covariate because its association with infection could influence the likelihood of receiving antibiotic treatment. Baseline demographics such as sex, age at diagnosis, US geographic census region, residential population density,36 Yost US-based socioeconomic status quintiles,37,38 tumor location, radiation therapy, and time from surgery to chemotherapy initiation were also used. The National Cancer Institute Comorbidity Index was included to adjust for differences in underlying health problems because it has demonstrated utility for site-specific cancers.39 Race from the Research Triangle Institute (RTI) race code was included to assess differences in distribution that may affect the results. Within the race category, American Indian/Alaska Native, Asian/Pacific Islander, Black (or African American), and Hispanic patients were grouped as “Other” because their individual cohort sizes were too small to show without potentially breaching patient anonymity according to SEER-Medicare standards.
One to one propensity score matching was performed between the antibiotic and no antibiotic groups (Figure S1), and Kaplan-Meier curves were generated to assess survival. Curves were generated for both the unmatched and matched cohorts, and results were further stratified by chemotherapy. Cox proportional hazards models were used to estimate the association between receiving antibiotics and overall survival, the association between antibiotic type received and overall survival in gemcitabine-treated patients, and the association between antibiotic receipt and cancer-specific survival. All hypothesis tests were 2 sided, with a statistical significance threshold set at P less than .05. Statistical analyses were conducted using the Python, version 3.11.5, programming language.
Results
A total of 712 patients with resected, early-stage pancreatic ductal adenocarcinoma treated with adjuvant chemotherapy were identified between January 1, 2007, and December 31, 2017 (Figure 1). Of these patients, 177 (24.9%) received antibiotics within the 1 month following chemotherapy initiation (Table 1). The mean (SD) age at diagnosis was 73.7 (5.1) years, and there were 402 (56.5%) women and 310 (43.5%) men. Most patients were White (582/712 [81.7%]), from a metropolitan area (613/712 [86.1%]), from the northeast or west US (490/712 [68.8%]), and belonged to the 2 highest socioeconomic quintiles (424/712 [59.6%]). For adjuvant first-line chemotherapy, the majority of patients received gemcitabine (629/712 [88.3%]), and few received fluorouracil (83/712 [11.7%]). Patients began chemotherapy an average (SD) of 7.52 (3.8) weeks after their surgery. Fewer than half of patients received endoscopic retrograde cholangiopancreatography (228/712 [32.0%]) or experienced any infection during the month following chemotherapy initiation (117/712 [16.4%]). Demographic and clinical characteristics between patients who did and did not receive antibiotics within 1 month of chemotherapy initiation were more balanced after propensity score matching, as demonstrated by the decrease in standard mean differences after matching (Table 1, Table S6). The overlap in distribution of propensity scores between patients treated with antibiotics and patients given no antibiotics indicates successful matching, allowing for meaningful comparisons between the 2 groups (Figure S1).
Table 1.
Demographic and clinical characteristics of patients with resected, early stage pancreatic ductal adenocarcinomaa
| Characteristic | Antibiotics (n = 177) | No antibiotics (n = 535) | Standardized mean difference |
|---|---|---|---|
| Sex, No. (%) | |||
| Female | 100 (56.5) | 302 (56.4) | .001 |
| Male | 77 (43.5) | 233 (43.6) | |
| Age at diagnosis, mean (SD), y | 73.8 (5.1) | 73.5 (5.1) | .06 |
| Radiation therapy, No. (%) | |||
| Yes | 29 (16.4) | 123 (23.0) | .17 |
| No | 148 (83.6) | 412 (77.0) | |
| Research Triangle Institute (RTI) race code, No. (%) | |||
| White | 153 (86.4) | 429 (80.2) | .17 |
| Otherb | 24 (13.6) | 106 (19.8) | |
| US Census region, No. (%) | |||
| West | 81 (45.8) | 231 (43.2) | .02 |
| Northeast | 29 (16.4) | 149 (27.8) | |
| South | 45 (25.4) | 91 (17.0) | |
| Midwest | 22 (12.4) | 64 (12.0) | |
| Population density, No. (%) | |||
| Metropolitan area | 147 (83.1) | 466 (87.1) | .11 |
| Nonmetropolitan area | 30 (16.9) | 69 (12.9) | |
| Yost Index, No. (%) | |||
| >80 to 100 | 55 (31.1) | 204 (38.1) | .10 |
| >60 to 80 | 38 (21.5) | 127 (23.7) | |
| >40 to 60 | 37 (20.9) | 66 (12.3) | |
| >20 to 40 | 29 (16.4) | 69 (12.9) | |
| >0 to 20 | 18 (10.1) | 69 (12.9) | |
| Year of diagnosis, No. (%) | |||
| 2013-2017 | 75 (42.4) | 221 (41.3) | .04 |
| 2010 to <2013 | 59 (33.3) | 171 (32.0) | |
| 2007 to <2010 | 43 (24.3) | 143 (26.7) | |
| Site of disease, No. (%) | |||
| Head | 162 (91.5) | 484 (90.5) | .06 |
| Body and neck/tail/unspecifiedc | 15 (8.5) | 51 (9.5) | |
| Surgery to chemotherapy start, mean (SD), wk | 7.71 (4.1) | 7.32 (3.4) | .10 |
| National Cancer Institute Comorbidity Index, No. (%) | |||
| 0 | 59 (33.3) | 224 (41.9) | .13 |
| >0, ≤ 1 | 99 (55.9) | 254 (47.5) | |
| >1 | 19 (10.7) | 57 (10.7) | |
| Endoscopic retrograde cholangiopancreatography, No. (%) | |||
| Yes | 59 (33.3) | 169 (31.6) | .04 |
| No | 118 (66.7) | 366 (68.4) | |
| Infection within 1 mo of chemotherapy start, No. (%) | |||
| Yes | 71 (40.1) | 46 (8.6) | .80 |
| No | 106 (59.9) | 489 (91.4) |
Demographic and clinical characteristics for the full cohort of patients, stratified by patients who received antibiotics in the month following chemotherapy (n = 177) and patients who did not receive antibiotics (n = 535).
Within the Race category, Black (or African American), Asian/Pacific Islander, Hispanic, American Indian/Alaska Native, and Other were combined into 1 category because their individual cohort sizes were too small to show without potentially breaching patient anonymity according to Surveillance, Epidemiology, and End Results Program–Medicare guidelines (n < 11) and in accordance with the Centers for Medicare & Medicaid Services cell size suppression policy.
Site of Disease at the Body and Neck, Tail, and Unspecified were combined into 1 category to comply with the Centers for Medicare & Medicaid Services cell size suppression policy.
Of the 143 propensity score–matched pairs, median survival among patients who received antibiotics during the 1 month following chemotherapy initiation was 23.9 months (95% CI = 20.4 to 27.1) compared with 18.2 months (95% CI = 15.9 to 20.9) for patients who did not receive antibiotics during this time period (P = .001) (Figure 2). When stratified by chemotherapy, median survival for the gemcitabine-treated patients who received antibiotics within the 1 month following chemotherapy was 24.3 months (95% CI = 20.8 to 27.9) compared with 20.0 months (95% CI = 17.3 to 21.6) for patients who did not receive antibiotics (P = .001). We were unable to determine association between antibiotic treatment and overall survival in patients who received adjuvant fluorouracil because there were only 12 propensity score–matched pairs. All but 32 patients had died by the end of study follow-up. Survival analyses before propensity score matching revealed similar results (Figure S2).
Figure 2.
Association of antibiotic receipt with overall survival in a propensity score–matched cohort. Five-year survival for patients who did and did not receive antibiotics during the exposure period, 1 month after initiation of chemotherapy. In the full propensity score–matched cohort (A), median survival with antibiotics was 23.9 months (95% CI = 20.4 to 27.1 months) and with no antibiotics was 18.2 months (95% CI = 15.9 to 20.9 months) (log rank P = .001). In the gemcitabine cohort (B), median survival with antibiotics was 24.3 months (95% CI = 20.8 to 27.9 months) and with no antibiotics was 20.0 months (95% CI = 17.3 to 21.6 months) (log rank P = .001). * Masking to protect patient anonymity due to small differences by subtraction to adjacent numbers at risk.
In the propensity score–matched Cox proportional hazards analyses, treatment with antibiotics during the 1 month after chemotherapy initiation was associated with improved overall survival (Table 2) (hazard ratio [HR] = 0.63, 95% CI = 0.49 to 0.81; P = .0004). On stratification by chemotherapy, we found that antibiotic treatment with gemcitabine was associated with an improved overall survival compared with treatment without antibiotics (HR = 0.70, 95% CI = 0.54 to 0.92; P = .009). Furthermore, when looking at cancer-specific survival, antibiotic treatment in the 1 month following chemotherapy was associated with improved cancer-specific survival (HR = 0.70, 95% CI = 0.53 to 0.92; P = .01). Similarly, antibiotic treatment was associated with a statistically significant improvement in cancer-specific survival among patients who received gemcitabine (HR = 0.73, 95% CI = 0.55 to 0.96; P = .03).
Table 2.
Overall survival vs cancer-specific survival associated with antibiotic receipt
| Antibiotic group outcome | Propensity score–matched hazard ratio (95% CI) | P |
|---|---|---|
| Overall survival | ||
| All chemotherapy | 0.63 (0.49 to 0.81) | .0004 |
| Gemcitabine | 0.70 (0.54 to 0.92) | .009 |
| Cancer-specific survival | ||
| All chemotherapy | 0.70 (0.53 to 0.92) | .01 |
| Gemcitabine | 0.73 (0.55 to 0.96) | .03 |
Among gemcitabine-treated patients who received antibiotic treatment, we wanted to determine whether treatment with a specific antibiotic coverage, class, or route of administration was associated with improved overall survival compared with treatment with any antibiotic. Among the 167 patients who received antibiotics, 93.4% received antibiotics with gram-negative coverage, which was not associated with a statistically significant change in overall survival (HR = 1.4, 95% CI = 0.61 to 3.17; P = .43). We did not see any association between overall survival and specific antibiotic class received (Table 3). Most antibiotics given were nonpenicillin β-lactams (47.3%) and quinolones (37.7%). Of the people who received antibiotics during the exposure window, 37.7% received at least 1 injected antibiotic, and 75.4% received at least 1 oral antibiotic. These subgroups were not associated with a change in overall survival.
Table 3.
Overall survival associated with antibiotic receipt, stratified by property among patients receiving gemcitabinea
| Antibiotic property | Sample size, No. (% of total) | Overall survival, hazard ratio (95% CI) | P |
|---|---|---|---|
| Coverage classb | |||
| Gram-negative | 156 (93.4) | 1.40 (0.61 to 3.17) | .43 |
| Non-penicillin β-lactamc | 79 (47.3) | 1.01 (0.47 to 2.13) | .99 |
| Quinolones | 63 (37.7) | 1.10 (0.55 to 2.20) | .79 |
| Penicillins | 19 (11.4) | 1.94 (0.91 to 4.16) | .09 |
| Macrolides | 14 (8.4) | 0.88 (0.35 to 2.18) | .78 |
| Other antibioticsd | 23 (13.8) | 0.91 (0.46 to 1.79) | .79 |
| Route | |||
| Injection | 63 (37.7) | 0.82 (0.36 to 1.83) | .62 |
| Oral | 126 (75.4) | 0.78 (0.38 to 2.18) | .83 |
Overall survival, stratified by antibiotic class or route for patients treated with adjuvant gemcitabine who received antibiotics within 1 month of chemotherapy initiation (n = 167). Antibiotic sample sizes and percentages do not sum to the total because some patients may have received more than 1 antibiotic during the 1 month.
Data are not shown for tetracyclines, sulfonamides, and aminoglycosides to comply with the Centers for Medicare & Medicaid Services cell size suppression policy.
Non-penicillin β-lactams include cephalosporins, carbapenems, and monobactam antibiotics.
Other antibiotics include lincosamide, glycopeptide, lipoglycopeptide, oxazolidinone, streptogramin, rifamycin, and nitroimidazole antibiotics.
We also performed sensitivity analyses, varying the antibiotic exposure window (Table S5). Improvement in overall survival was noted only when antibiotic exposure took place within the 1 month after chemotherapy initiation (HR = 0.63, 95% CI = 0.49 to 0.81; P = .0004). There was no association between antibiotic treatment and overall survival when the antibiotic exposure window took place between a month before chemotherapy initiation and a month after chemotherapy initiation (HR = 0.95, 95% CI = 0.79 to 1.15; P = .60) or when it took place within 3 months after chemotherapy initiation (HR = 0.89, 95% CI = 0.73 to 1.08; P = .27), as presented in Table S5.
Discussion
In this retrospective cohort study involving SEER-Medicare patients with resected, early-stage pancreatic ductal adenocarcinoma, antibiotic treatment within 1 month after chemotherapy initiation was associated with improved overall survival and improved cancer-specific survival. This improvement was particularly notable among patients receiving gemcitabine. A prior study found that a majority (52%) of the pancreatic ductal adenocarcinoma microbiome is composed of Gammaproteobacteria, a bacteria that produces cytidine deaminase capable of deaminating gemcitabine (2′,2′-difluorodeoxycytidine) into its inactive form (2′,2′-difluorodeoxyuridine).13 The study showed that tumor-bearing mice injected with Escherichia coli–expressing cytidine deaminase had enhanced antitumor response when treated with gemcitabine and ciprofloxacin compared with gemcitabine alone. Furthermore, bacteria obtained from human pancreatic ductal adenocarcinoma samples were shown to mediate gemcitabine resistance in colon cancer cell lines.13 Altogether, this research suggests that Gammaproteobacteria within the pancreatic ductal adenocarcinoma microbiome may mediate gemcitabine resistance through cytidine deaminase production. Antibiotic treatment could offer a strategy to reduce Gammaproteobacteria in the pancreatic microbiome, limiting their ability to inactivate gemcitabine and thereby enhancing chemoeffectiveness.
In a multicenter retrospective study of 211 patients with pancreatic ductal adenocarcinoma undergoing pancreaticoduodenectomy, adjuvant gemcitabine chemotherapy was associated with improved progression-free survival in patients with a negative intraoperative bile culture for Klebsiella pneumoniae compared to patients who had a positive bile culture.17 As a result, the potential impact that K pneumoniae, a subspecies of Gammaproteobacteria, has on gemcitabine efficacy merits further investigation in prospective studies. Our results underscore this existing literature and support the hypothesis that use of antibiotics may have a role in mitigating the negative metabolic impact the pancreatic ductal adenocarcinoma microbiome has on gemcitabine efficacy.
A prior study from our group investigated antibiotic treatment in patients with metastatic pancreatic ductal adenocarcinoma within a perichemotherapy window, defined as antibiotic exposure occurring 1 month before and 1 month after initiation of chemotherapy.19 In this current study, we observed an association with improvement in overall survival only when antibiotic exposure took place within the first month after chemotherapy initiation. There was no statistically significant change in overall survival when the exposure window was defined as 1 month before and 1 month after chemotherapy initiation. Treatment with antibiotics within 1 month of chemotherapy initiation represents an even more immediate time frame, highlighting the possible temporal impact of antibiotic treatment when starting chemotherapy. Our findings support the hypothesis that closely aligning antibiotic exposure with chemotherapy initiation may help mitigate bacteria-mediated resistance, specifically in patients treated with gemcitabine.
Finally, we aimed to determine whether treatment with specific antibiotic types is associated with overall survival in patients receiving gemcitabine. Because Gammaproteobacteria are gram negative, we hypothesized that gram-negative antibiotic treatment may be associated with improved survival compared with antibiotic treatment without gram-negative coverage. Among gemcitabine-treated patients given antibiotic treatment, however, 93.4% of the antibiotics had gram-negative coverage. As a result, we were unable to determine survival differences based on coverage specificity. In addition, stratification by antibiotic class showed no statistically significant survival differences, likely because of a small patient cohort made even smaller after stratification.
In this study, various measures were taken to validate and strengthen our findings. To address confounding bias and improve comparability between the antibiotic and no-antibiotic groups, we performed propensity score estimation and matching. By doing so, we balanced covariate distribution between groups. We aimed to eliminate survivorship bias and prevent overly optimistic results by confining the antibiotic treatment window to 1 month after chemotherapy initiation. Finally, we believe that the associated survival benefit from antibiotic treatment may be an underestimation because patients typically receive antibiotics when they are ill or in a compromised health state. As a result, the apparent survival advantage within this group could be underestimated, given that these patients were likely unwell from the outset.
Despite the many strengths in this study, there were several limitations. Patients were primarily older, White, from metropolitan areas in the northeastern or western US, and affluent. The majority of pancreatic ductal adenocarcinoma tumors were in the pancreatic head, with few in the body or tail. The homogeneous nature of the patient population limits our study’s generalizability because the broader population exhibits greater demographic and clinical variability. In addition, the unknown state of each patient’s microbiome may have influenced survival, and in this claims-based study we could not assess whether antibiotic treatment altered the pancreatic ductal adenocarcinoma microbiome or chemotherapy efficacy. We were unable to link antibiotic prescription claims to medical claims in the Medicare database; as a result, we could not identify indications for antibiotics, which could also have influenced survival. Infection during the time of antibiotic use, however, was included as a covariate to address real-time differences in health status.
Our small sample size limited the breadth of our analysis. Propensity score matching contributed to the reduction in the sample size, which, in turn, may limit the external validity of our results. Other propensity score matching limitations include confounding because the method cannot adjust for unmeasured factors that may have influenced the indication for antibiotic treatment as well as conditional independence. Furthermore, clinical trials have recently suggested a benefit to neoadjuvant therapy for resected, early-stage pancreatic ductal adenocarcinoma40,41; however, few patients received neoadjuvant chemotherapy during the study period and therefore were not included in our cohort. We also emphasize that causation can in no way be proved because this is a retrospective cohort study. We observed an association between antibiotic treatment and improved overall survival, but we are unable to determine whether antibiotic treatment was directly responsible for the observed survival benefit. Finally, use of antibiotics may be associated with improved survival by mechanisms unrelated to their antimicrobial properties. For example, patients who received antibiotics may have had more intensive follow-up after surgical resection of pancreatic ductal adenocarcinoma that therefore improved their overall survival. Future prospective studies investigating the effects of antibiotic treatment on survival, combined with monitoring changes in the pancreatic ductal adenocarcinoma microbiome, are warranted.
Our findings demonstrate that antibiotic treatment within 1 month after gemcitabine initiation is associated with improved survival, highlighting a potential role for antibiotic treatment in improving chemotherapeutic efficacy and pancreatic ductal adenocarcinoma outcomes. Future studies may evaluate specific antibiotic coverage responsible for the observed survival benefit and provide a more comprehensive understanding of the pancreatic ductal adenocarcinoma microbiome. We hope that future investigations can be conducted within a more diverse patient population to provide comprehensive insights accessible to individuals of all backgrounds.
Supplementary Material
Acknowledgments
We thank the Digestive Disease Research Foundation for supporting this research. We greatly appreciate the efforts of the National Cancer Institute, Information Management Services Inc, and the SEER program tumor registries in the creation of the SEER-Medicare database. To disclaim, the ideas and opinions expressed herein are those of the authors and do not necessarily reflect the opinions of the state of California, Department of Public Health, the National Cancer Institute, or the Centers for Disease Control and Prevention or their contractors and subcontractors.
Contributor Information
Emma Gong, Henry D. Janowitz Division of Gastroenterology, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States.
Daniel J Fulop, Division of General Internal Medicine, Massachusetts General Hospital, Boston, MA 02114, United States.
Joyce Serebrenik, Henry D. Janowitz Division of Gastroenterology, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States.
Arielle J Labiner, George Washington University School of Medicine and Health Sciences, Washington, DC 20052, United States.
Deirdre J Cohen, Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States.
Keith M Sigel, Division of General Internal Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States.
Aimee L Lucas, Henry D. Janowitz Division of Gastroenterology, Icahn School of Medicine at Mount Sinai, New York, NY 10029, United States.
Author contributions
Emma Wai Man Gong, BS (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Writing—original draft, Writing—review & editing), Daniel J. Fulop, MD, MS (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—review & editing), Joyce Serebrenik, BA, MS (Project administration, Writing—review & editing), Arielle Labiner, BA, MHS (Writing—review & editing), Deirdre Cohen, MD, MS (Conceptualization, Data curation, Formal analysis, Supervision, Writing—review & editing), Keith M. Sigel, MD, PhD (Conceptualization, Data curation, Formal analysis, Supervision, Writing—review & editing), Aimee Lucas, MD, MS (Conceptualization, Data curation, Formal analysis, Project administration, Supervision, Validation, Writing—review & editing).
Supplementary material
Supplementary material is available at JNCI Cancer Spectrum online.
Funding
The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code section 103885; the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; the National Cancer Institute’s SEER Program under contract No. HHSN261201800032I awarded to the University of California, San Francisco, contract No. HHSN261201800015I awarded to the University of Southern California, and contract No. HHSN261201800009I awarded to the Public Health Institute. The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Conflicts of interest
Ms Gong reported receiving grants from the Digestive Disease Research Foundation during the conduct of the study. Dr Cohen reported receiving grants and consulting fees from Astellas Pharma US, Inc; Guardant; Loxo@Lilly; and Seagen outside the submitted work. Dr Lucas reported receiving grants and consulting fees from Immunovia, Aionco, and ClearNote Health as well as royalties from UpToDate outside the submitted work. No other disclosures were reported.
Data availability
The SEER-Medicare data are available to investigators for research purposes. Maintaining patient and clinician confidentiality is a primary concern of the National Cancer Institute, SEER, and the Centers for Medicare & Medicaid Services. Therefore, SEER-Medicare data are not public use data files. More information can be found at https://healthcaredelivery.cancer.gov/seermedicare/obtain/.
References
- 1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74:12-49. [DOI] [PubMed] [Google Scholar]
- 2. Arnold M, Abnet CC, Neale RE, et al. Global burden of 5 major types of gastrointestinal cancer. Gastroenterology. 2020;159:335-349 e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. 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. [DOI] [PubMed] [Google Scholar]
- 4. 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]
- 5. Conroy T, Hammel P, Hebbar M, et al. ; Canadian Cancer Trials Group and the Unicancer-GI–PRODIGE Group. FOLFIRINOX or gemcitabine as adjuvant therapy for pancreatic cancer. N Engl J Med. 2018;379:2395-2406. [DOI] [PubMed] [Google Scholar]
- 6. Park W, Chawla A, O'Reilly EM. Pancreatic cancer: a review. JAMA. 2021;326:851-862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Pushalkar S, Hundeyin M, Daley D, et al. The pancreatic cancer microbiome promotes oncogenesis by induction of innate and adaptive immune suppression. Cancer Discov. 2018;8:403-416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Nilsson HO, Stenram U, Ihse I, Wadstrom T. Helicobacter species ribosomal DNA in the pancreas, stomach and duodenum of pancreatic cancer patients. World J Gastroenterol. 2006;12:3038-3043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Mitsuhashi K, Nosho K, Sukawa Y, et al. Association of Fusobacterium species in pancreatic cancer tissues with molecular features and prognosis. Oncotarget., 2015;6:7209-7220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Riquelme E, Zhang Y, Zhang L, et al. Tumor microbiome diversity and composition influence pancreatic cancer outcomes. Cell. 2019;178:795-806 e12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Capula M, Perán M, Xu G, et al. Role of drug catabolism, modulation of oncogenic signaling and tumor microenvironment in microbe-mediated pancreatic cancer chemoresistance. Drug Resist Updat. 2022;64:100864. [DOI] [PubMed] [Google Scholar]
- 12. Lehouritis P, Cummins J, Stanton M, et al. Local bacteria affect the efficacy of chemotherapeutic drugs. Sci Rep. 2015;5:14554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Geller LT, Barzily-Rokni M, Danino T, et al. Potential role of intratumor bacteria in mediating tumor resistance to the chemotherapeutic drug gemcitabine. Science. 2017;357:1156-1160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Geller LT, Straussman R. Intratumoral bacteria may elicit chemoresistance by metabolizing anticancer agents. Mol Cell Oncol. 2018;5:e1405139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Vande Voorde J, Sabuncuoğlu S, Noppen S, et al. Nucleoside-catabolizing enzymes in mycoplasma-infected tumor cell cultures compromise the cytostatic activity of the anticancer drug gemcitabine. J Biol Chem. 2014;289:13054-13065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Chrysostomou D, Roberts LA, Marchesi JR, Kinross JM. Gut microbiota modulation of efficacy and toxicity of cancer chemotherapy and immunotherapy. Gastroenterology. 2023;164:198-213. [DOI] [PubMed] [Google Scholar]
- 17. Weniger M, Hank T, Qadan M, et al. Influence of Klebsiella pneumoniae and quinolone treatment on prognosis in patients with pancreatic cancer. Br J Surg. 2021;108:709-716. [DOI] [PubMed] [Google Scholar]
- 18. Mohindroo C, Hasanov M, Rogers JE, et al. Antibiotic use influences outcomes in advanced pancreatic adenocarcinoma patients. Cancer Med. 2021;10:5041-5050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Fulop DJ, Zylberberg HM, Wu YL, et al. Association of antibiotic receipt with survival among patients with metastatic pancreatic ductal adenocarcinoma receiving chemotherapy. JAMA Netw Open. 2023;6:e234254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Nakano S, Komatsu Y, Kawamoto Y, et al. Association between the use of antibiotics and efficacy of gemcitabine plus nab-paclitaxel in advanced pancreatic cancer. Medicine (Baltimore). 2020;99:e22250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Park HS, Lloyd S, Decker RH, et al. Overview of the Surveillance, Epidemiology, and End Results database: Evolution, data variables, and quality assurance. Curr Probl Cancer. 2012;36:183-190. [DOI] [PubMed] [Google Scholar]
- 22. Nipp R, Tramontano AC, Kong CY, et al. Disparities in cancer outcomes across age, sex, and race/ethnicity among patients with pancreatic cancer. Cancer Med. 2018;7:525-535. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Zylberberg HM, Woodrell C, Rustgi SD, et al. Opioid prescription is associated with increased survival in older adult patients with pancreatic cancer in the united states: a propensity score analysis. J Clin Oncol Oncol Pract. 2022;18:e659-e668. [DOI] [PubMed] [Google Scholar]
- 24. E J-Y, Lu S-E, Lin Y, et al. Differential and joint effects of metformin and statins on overall survival of elderly patients with pancreatic adenocarcinoma: a large population-based study. Cancer Epidemiol Biomarkers Prev. 2017;26:1225-1232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Bilimoria KY, Bentrem DJ, Ko CY, et al. Validation of the 6th edition AJCC pancreatic cancer staging system: report from the National Cancer Database. Cancer. 2007;110:738-744. [DOI] [PubMed] [Google Scholar]
- 26. Azap RA, Diaz A, Hyer JM, et al. Impact of race/ethnicity and county-level vulnerability on receipt of surgery among older Medicare beneficiaries with the diagnosis of early pancreatic cancer. Ann Surg Oncol. 2021;28:6309-6316. [DOI] [PubMed] [Google Scholar]
- 27. Valle JW, Palmer D, Jackson R, et al. Optimal duration and timing of adjuvant chemotherapy after definitive surgery for ductal adenocarcinoma of the pancreas: ongoing lessons from the ESPAC-3 study. J Clin Oncol. 2014;32:504-512. [DOI] [PubMed] [Google Scholar]
- 28. Wood LD, Canto MI, Jaffee EM, Simeone DM. Pancreatic cancer: pathogenesis, screening, diagnosis, and treatment. Gastroenterology. 2022;163:386-402.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Turpin A, El Amrani M, Bachet J-B, et al. Adjuvant pancreatic cancer management: towards new perspectives in 2021. Cancers (Basel). 2020;12: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. 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;24:2770-2776. [DOI] [PubMed] [Google Scholar]
- 31. Turner MC, Masoud SJ, Cerullo M, et al. Improved overall survival is still observed in patients receiving delayed adjuvant chemotherapy after pancreaticoduodenectomy for pancreatic adenocarcinoma. HPB (Oxford). 2020;22:1542-1548. [DOI] [PubMed] [Google Scholar]
- 32. Manichanh C, Reeder J, Gibert P, et al. Reshaping the gut microbiome with bacterial transplantation and antibiotic intake. Genome Res. 2010;20:1411-1419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Derosa L, Hellmann MD, Spaziano M, et al. Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer. Ann Oncol, 2018;29:1437-1444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science. 2018;359:91-97. [DOI] [PubMed] [Google Scholar]
- 35. Sen S, Carmagnani Pestana R, Hess K, et al. Impact of antibiotic use on survival in patients with advanced cancers treated on immune checkpoint inhibitor phase I clinical trials. Ann Oncol. 2018;29:2396-2398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Moss JL, Stinchcomb DG, Yu M. Providing Higher Resolution Indicators of Rurality in the Surveillance, Epidemiology, and End Results (SEER) Database: Implications for Patient Privacy and Research. Cancer Epidemiol Biomarkers Prev. 2019;28:1409-1416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Yost K, Perkins C, Cohen R, et al. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control. 2001;12:703-711. [DOI] [PubMed] [Google Scholar]
- 38. Yu M, Tatalovich Z, Gibson JT, Cronin KA. Using a composite index of socioeconomic status to investigate health disparities while protecting the confidentiality of cancer registry data. Cancer Causes Control. 2014;25:81-92. [DOI] [PubMed] [Google Scholar]
- 39. Klabunde CN, Legler JM, Warren JL, et al. A refined comorbidity measurement algorithm for claims-based studies of breast, prostate, colorectal, and lung cancer patients. Ann Epidemiol. 2007;17:584-590. [DOI] [PubMed] [Google Scholar]
- 40. Versteijne E, van Dam JL, Suker M, et al. ; Dutch Pancreatic Cancer Group., Neoadjuvant chemoradiotherapy versus upfront surgery for resectable and borderline resectable pancreatic cancer: long-term results of the Dutch randomized PREOPANC trial. J Clin Oncol. 2022;40:1220-1230. [DOI] [PubMed] [Google Scholar]
- 41. Nassoiy S, Christopher W, Marcus R, et al. Evolving management of early stage pancreatic adenocarcinoma in older patients. Am J Surg. 2023;225:212-219. [DOI] [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 SEER-Medicare data are available to investigators for research purposes. Maintaining patient and clinician confidentiality is a primary concern of the National Cancer Institute, SEER, and the Centers for Medicare & Medicaid Services. Therefore, SEER-Medicare data are not public use data files. More information can be found at https://healthcaredelivery.cancer.gov/seermedicare/obtain/.