Pancreatic Cancer Patients Discover Weight Measurements Can Be Deceptive
Overview
Elevated body mass index (BMI) has been associated with an increased risk of pancreatic cancer (PC). However, cancer-associated weight loss occurring years before diagnosis may have led to an underestimation of this association. The extent to which this potential bias has been considered in past epidemiological research remains unclear. To address this, a systematic search was conducted in two databases up to February 19, 2024, to identify systematic reviews, meta-analyses, and pooled analyses examining the relationship between BMI and PC. The focus was on whether these studies accounted for prediagnostic weight loss as a source of bias and how included cohort studies handled this concern.
A total of thirteen relevant review articles, meta-analyses, and pooled analyses were identified. Of these, only five studies (comprising four pooled analyses and one systematic review) took prediagnostic weight loss into account. Furthermore, 24 out of 32 identified cohort studies reported excluding the initial years of follow-up to address potential bias, but only 13 studies provided results after such exclusions. Notably, the effect estimates for the association between BMI and PC risk generally increased with longer exclusion periods of initial follow-up.
This analysis suggests that the relationship between overweight/obesity and the risk of pancreatic cancer is likely stronger than previously reported. Future research should diligently address and mitigate potential bias arising from prediagnostic weight loss to provide more accurate estimates of this association.
Introduction
Pancreatic cancer (PC) is a particularly deadly disease, with a dismal five-year survival rate of approximately 12% in the United States. The incidence of PC has been rising in various regions, including the U.S., and it is anticipated to become the second leading cause of cancer-related deaths by 2026, overtaking colorectal cancer. This alarming trend underscores the critical need for understanding and addressing risk factors associated with pancreatic cancer.
One significant risk factor is elevated body mass index (BMI), which categorizes individuals as overweight (BMI ≥25 to <30 kg/m2) or obese (BMI ≥30 kg/m2). According to the World Cancer Research Fund (WCRF), there is robust evidence linking higher BMI with an increased risk of developing pancreatic cancer. This is further supported by previous systematic reviews, which have consistently reported that the risk of PC rises by 8–12% for every five-unit increase in BMI.
A complicating factor in studying the relationship between BMI and pancreatic cancer risk is the phenomenon of unintentional prediagnostic weight loss, which is prevalent among PC patients. Studies indicate that this type of weight loss occurs in 50%–75% of PC cases and can manifest long before a formal diagnosis is made. This weight loss is typically characterized as a reduction in body weight by more than 5% over a six-month period. The potential for cancer-associated weight change to begin months or even years before diagnosis poses a significant challenge for accurately determining the impact of BMI on pancreatic cancer risk.
This issue is particularly relevant in case–control studies, where BMI is often measured close to the time of diagnosis, potentially after significant cancer-related weight loss has already occurred. In such cases, BMI measurements may not accurately reflect the individual’s typical body weight prior to the onset of cancer, thus biasing the results. Similarly, in cohort studies, the BMI recorded at baseline may not accurately represent an individual’s usual body weight if pancreatic cancer was already developing in a preclinical state at the time of enrollment. Consequently, cases diagnosed early in the follow-up period may already be experiencing weight loss due to undiagnosed cancer, which can lead to an underestimation of the true association between BMI and pancreatic cancer risk.
To explore how previous research has addressed these issues, an umbrella review was conducted, examining systematic reviews, meta-analyses, and pooled analyses that investigated the BMI-PC association. This review specifically focused on how these studies considered the potential bias introduced by prediagnostic weight loss. It also assessed how individual cohort studies included in these reviews managed the potential confounding effect of weight loss due to undiagnosed cancer.
Also read; Pancreatic Cancer Danger In Diabetic Women
The findings from this investigation revealed that not all studies adequately addressed the issue of prediagnostic weight loss. In case–control studies, it is crucial to account for weight before any potential cancer-related weight loss, while in cohort studies, careful consideration must be given to cases diagnosed during the initial years of follow-up. Failure to do so can result in biased estimates, underestimating the true impact of BMI on pancreatic cancer risk. Therefore, future epidemiological studies must incorporate strategies to minimize this bias, ensuring more accurate assessments of the relationship between BMI and the risk of developing pancreatic cancer.
Methods
Inclusion Criteria:
- Study Types: The study included systematic reviews, meta-analyses, and pooled analyses.
- Exposure of Interest: The primary exposure of interest was obesity or overweight, defined by Body Mass Index (BMI), or excess weight per unit increase in BMI.
- Outcome: The primary outcome of interest was the development of pancreatic cancer (PC).
Exclusion Criteria:
- Non-BMI Measures: Studies that exclusively used measures of adiposity other than BMI (such as waist-to-hip ratio, waist circumference, or weight gain) were excluded from the analysis.
The study protocol was registered with PROSPERO under registration number CRD42022333665 before initiating the study. Adhering to standardized methodology guidelines, the research followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework. A systematic search was conducted using predefined algorithms in PubMed and Web of Science, covering the period until February 19, 2024. The search targeted systematic reviews, meta-analyses, and pooled analyses that explored the relationship between overweight and obesity and the risk of developing pancreatic cancer. Additionally, references from the included articles were manually reviewed to identify any missed studies.
Two reviewers independently extracted data from each study, including the first author’s name, article title, publication year, number of studies included, exposure definition, summary effect size estimates (using the most adjusted estimates), confidence intervals (CIs), model types, and measures of heterogeneity. If available, a combined effect estimate for both genders was recorded; otherwise, separate effect size estimates for men and women were reported. Discrepancies between the reviewers were resolved through discussion or consultation with a third reviewer.
The methodological quality of the included systematic reviews was assessed using the AMSTAR-2 tool, which evaluates the quality of systematic reviews. In line with AMSTAR-2 guidelines, an overall score for individual studies was not provided.
Potential bias from prediagnostic weight loss was evaluated using two criteria: “timing of BMI ascertainment” for case–control studies and “consideration of sojourn time” for cohort studies. A review or meta-analysis was considered to have addressed “timing of BMI ascertainment” if the exact timing of BMI measurement was reported for each primary study and was accounted for in the analysis, either through stratification, subgroup analysis, or the exclusion of studies where BMI was measured within one year of diagnosis. The “consideration of sojourn time” criterion was deemed met if analyses excluded at least the first year of follow-up in cohort studies.
To further investigate potential bias due to prediagnostic weight loss, information from individual studies included in the 2012 WCRF/AICR systematic review, which focused on cohort studies, was extracted and updated with studies published between 2011 and 2024. Hazard ratios (HRs) and their CIs from primary analyses were compared with those from sensitivity analyses in studies that extended the exclusion of initial follow-up years. All analyses and visualizations were conducted using R version 4.2.2.
Results
The systematic search identified 4,046 records, from which 13 publications met the inclusion criteria (see Figure 1). A detailed overview of the characteristics of these systematic reviews, meta-analyses, and pooled analyses is presented in Table 1. Among these, nine studies focused solely on cohort studies (either prospective or retrospective), while four studies included both cohort and case-control studies. The studies typically reported summary relative risk estimates comparing overweight and obesity with normal weight using the World Health Organization (WHO) BMI classifications (underweight: <18.5; normal weight: 18.5 to <25; overweight: 25 to <30; obesity: ≥30) or per 5-unit increase in BMI. For men, the summary relative risk estimates for overweight, obesity, and a 5-unit increase in BMI ranged from 1.13 to 1.28, 1.28 to 2.29, and 1.07 to 1.10, respectively. For women, these estimates ranged from 1.12 to 1.24, 1.28 to 1.60, and 1.08 to 1.12.
An evaluation of the methodological quality of the nine systematic reviews (excluding the four publications that were not systematic reviews) indicated concerns, particularly in items 9, 12, and 13 of the AMSTAR-2 tool, which assess the adequacy of techniques used to evaluate the risk of bias in the individual studies and the consideration of this bias in the meta-analysis results. Out of the 13 studies, five publications addressed the potential bias due to cancer-associated prediagnostic weight loss, while the others did not sufficiently consider this issue.
Reviews Addressing Prediagnostic Weight Loss Bias
- Koyanagi et al. conducted a pooled analysis of nine population-based cohort studies from Japan, examining the association between BMI and pancreatic cancer (PC) risk. Excluding cases diagnosed within the first three years of follow-up led to slightly increased hazard ratios (HRs), though the number of cases was limited, resulting in wide confidence intervals that included the null value.
- Aune et al. systematically reviewed cohort studies on the BMI-PC association and specifically addressed the handling of potential prediagnostic weight loss bias. They stratified analyses based on the exclusion of early follow-up periods. The summary relative risk (RR) for a 5-unit BMI increase was 1.10 (95% CI 1.07–1.14) when all studies were included, 1.11 (95% CI 1.05–1.18) when the first year was excluded, and 1.13 (95% CI 1.05–1.21) when the first two years were excluded.
- Arslan et al. analyzed one case-control and 11 cohort studies, excluding cases diagnosed in the first two years of follow-up. After such exclusion, there was a moderate increase in odds ratios (ORs) for all BMI categories: overweight (1.15 to 1.22), obesity class I (1.13 to 1.22), and obesity class II (1.26 to 1.32).
- Jiao et al. pooled seven prospective Japanese cohorts, calculating follow-up time from one year after baseline. The summary RR for a 5-unit increase in BMI was 1.06 (95% CI 0.99–1.13) for men and 1.12 (95% CI 1.05–1.19) for women.
- Genkinger et al. pooled 14 Japanese cohorts but did not report specific results from sensitivity analyses that excluded the first two or five years of follow-up.
Analysis of Individual Cohort Studies
Building upon Aune et al.’s systematic review, seven additional publications were identified, resulting in a total of 32 studies examining the BMI-PC association. Among these, 24 studies reported excluding initial years of follow-up, with only 13 providing results post-exclusion. Exclusions ranged from the first year to the first five years of follow-up. Sensitivity analyses generally showed higher estimates than main analyses, particularly for BMI in the obesity range.
Further analyses for a 5-unit increase in BMI, stratified by the exclusion of early follow-up years, yielded similar results to the 2012 review. Without exclusion, the summary RR was 1.08 (95% CI 1.06–1.11; I² = 21%, 24 studies). Excluding at least the first year resulted in an RR of 1.07 (95% CI 1.03–1.11; I² = 75%, nine studies), and excluding at least the first two years slightly increased the estimate to 1.08 (95% CI 1.03–1.14; I² = 58%, six studies).
Conclusion
This umbrella review is the first comprehensive evaluation of how previous systematic reviews, meta-analyses, and pooled analyses—along with their included primary studies—have considered prediagnostic weight loss as a potential source of bias in assessing the relationship between body mass index (BMI) and pancreatic cancer (PC). Our findings emphasize the critical need to thoroughly address this bias to accurately determine the impact of overweight and obesity on PC risk.
It is well-established that excess weight is positively associated with an increased risk of PC. However, many patients with pancreatic cancer experience unintentional weight loss before diagnosis, known as cancer cachexia. This condition is often defined as a loss of 5% of body weight within six months. Studies have indicated that weight loss in pancreatic cancer patients can commence years before diagnosis. For instance, a 2024 study by Wang et al. found that individuals who experienced a 5.1–10.0% weight loss had a 91% increased risk of developing pancreatic cancer, while those with greater than 10% weight loss faced a 328% increased risk compared to those with no weight loss. Additionally, a large nested case-control study by Tan et al., involving 28,137 PC cases and 261,219 controls, demonstrated a U-shaped relationship between BMI and pancreatic cancer. The study showed that a decline in BMI among pancreatic cancer cases began as early as 33 months prior to diagnosis, with an initial gradual decline phase followed by a rapid decline phase starting approximately nine months before diagnosis.
Oncology Related Tools
Other
Latest Research
