Screening for pancreatic cancer among smokers as high-risk individuals: Systematic review and meta-analysis of prospective cohort studies

• There are no formal screening guidelines for pancreatic cancer and often diagnosed late, with a high mortality rate. • This manuscript provides clinically relevant information for health care providers who manage this patient population. • This research is unique given its focus on asymptomatic versus hereditary diagnosis, using a robust search strategy. • This study is innovative for the following reasons: (1) use of the IVhet model over the less robust random-effects and fixed-effect(s) models that are used, (2) use of what is considered to be the more interpretable DOI plot versus the traditionally used funnel plot for small-study effects, (3) use of what is the more robust LFK index versus the traditionally used Egger's regression intercept test for the quantitative assessment of small-study, (4) use of 95 % prediction intervals, a statistic that is rarely used but a better measure of between-study heterogeneity than Q and I-squared. The main objective of this study was to conduct an aggregate data systematic review with meta-analysis of prospective cohort studies to determine the association between imaging screening for smoking and early diagnosis of pancreatic cancer (PC). Eligible studies included the following: 1) prospective cohort studies conducted in humans, 2) appropriate imaging screening method, 3) participants categorized into smoking as a high-risk group with no reported genetic disposition for PC, 4) no prior diagnosis of PC, 5) studies which detected and diagnosed adenocarcinoma following baseline screening, 6) data available for our primary outcome, PC, 7) studies published in the English language from January 1, 1992 to July 22, 2024. Any studies not meeting all of the above criteria were excluded. Information sources included the following: 1) PubMed, 2) CINAHL, 3) Scopus, 4) Cochrane Central Register of Controlled Trials (CENTRAL), 5) NIH National Cancer Institute's Division of Cancer Prevention, 6) Pancreatic Cancer Action Network, 7) ProQuest, 8) The British Medical Journal's Gut and Pancreatology , 9) Clinicaltrials.gov. The date of the last search (PubMed) was conducted on July 22, 2024. Risk of bias was assessed using the Cochrane Collaborations Risk Of Bias In Non-randomized Studies of Exposures [ 2 ] instrument. Small-study effects (publication bias, etc.) was assessed using the Doi plot and LFK index. The effect size metric for the primary outcome, PC, was the odds ratio (OR). Odd's ratios were pooled using the inverse-variance (IVhet) model. The strength/certainty of evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) instrument. Five retrospective cohort studies representing 3,517 patients met the criteria for inclusion, 944 of whom were smokers. Two of the five studies reported the exact diagnosis stage for smokers. Among cases of pancreatic cancer ( n = 34), the estimated effect of smoking suggested harm (OR = 2.24, 95 % CI, 0.96 to 5.23) though with weak statistical significance ( p = 0.06). Statistically significant heterogeneity ( Q = 11.1, p = 0.03) and moderate inconsistency ( I 2 = 63.9 %, 95 % CI, 5.0 to 86.3 %) were observed. The 95 % prediction interval (PI) was 0.15 to 32.38. Risk of bias, based on ROBINS-E ranged from high to very high, with very low certainty of evidence based on the GRADE assessment. While not statistically significant, our findings suggest that smokers may be at an increased odds of having pancreatic cancer. However, these findings need to be interpreted with respect to the high risk of bias and very low certainty of evidence observed. A need exists for additional, well-designed studies that include intentional assessment of screening methods for smoking and other high-risk groups to support clinical guideline development before any definitive conclusions can be drawn. No funding was received for this work. An a priori protocol was developed but not registered in any registry because of concerns about plagiarism.