Early‐life body size and risk of developing biliary tract cancers

Abstract

Background

Childhood body mass index (BMI) trajectories, BMI, height, and birth weight were investigated in relation to biliary tract cancer (BTC) risk in this population‐based cohort study.

Methods

The study included 172,113 males and 168,503 females born between 1930 and 1996 from the Copenhagen School Health Records Register. Heights and weights measured at ages 6–15 years identified five sex‐specific BMI trajectories. BMI and height were analyzed as z scores; overweight was defined via US Centers for Disease Control and Prevention criteria. Sex‐specific hazard ratios (HRs) were estimated via birth cohort–stratified Cox regressions.

Results

During a median follow‐up of 34.5 years, 635 individuals developed BTCs. Overweight (HR, 1.58; 95% confidence interval [CI], 1.06–2.34) and obesity trajectories in males (HR, 3.22; 95% CI, 1.61–6.44) and the obesity trajectory in females (HR, 2.88; 95% CI, 1.62–5.15) were associated with increased BTC risk compared with the average BMI trajectory. Childhood overweight at age 7 years was associated with increased intrahepatic bile duct cancer risk in males (HR, 2.66; 95% CI, 1.48–4.75) and extrahepatic bile duct cancer risk in females (HR, 3.83; 95% CI, 1.94–7.56). Taller childhood height was linked to a higher BTC risk in males only; birth weight showed no associations.

Conclusions

Childhood overweight and obesity increase BTC risk in adulthood.

This visual abstract illustrates how childhood body size influences biliary tract cancer (BTC) risk. By using more than 340,000 Danish children’s health records, body mass index trajectories between ages 6 and 15 years were investigated. The results show that obesity during childhood significantly increases BTC risk, with sex‐specific differences across intrahepatic, extrahepatic, gallbladder, and ampullary cancers.

INTRODUCTION

Biliary tract cancers (BTCs), which comprise gallbladder cancer (GBC), intrahepatic bile duct cancer (IHBDC), extrahepatic bile duct cancer (EHBDC), and ampulla of Vater cancer (AVC), are a group of rare but highly aggressive malignancies. ¹ , ² These cancers are characterized by poor prognosis, largely due to late‐stage diagnosis and limited treatment options. The distribution of BTCs varies considerably by geography and sex, with GBC more commonly diagnosed in females and IHBDC and EHBDC more prevalent in males. ¹ , ³ , ⁴ In the United States, the overall 5‐year survival rate for BTCs remains low at 15%, which underscores the critical need for better understanding of their etiology and the development of effective prevention strategies. ⁵

The role of metabolic factors, particularly obesity, in the pathogenesis of BTCs has gained increasing attention. ³ , ⁵ , ⁶ Obesity, now recognized as a chronic inflammatory state, ⁷ , ⁸ is strongly linked to an increased risk of all BTCs except for AVC, which makes it a potentially modifiable risk factor. ⁹ , ¹⁰ Although adult obesity has been extensively linked to metabolic disorders, type 2 diabetes, and various adult‐onset cancers, ⁸ , ⁹ , ¹¹ the existing literature has largely overlooked the role of childhood obesity in BTC etiology, which suggests a gap in understanding its potential early‐life effects. This gap is important, given the rising prevalence of obesity from an early age and its implications for long‐term health. ¹² , ¹³

Prior research indicates that exposure to excess weight during childhood is relevant for later obesity‐related cancers, ¹³ , ¹⁴ , ¹⁵ yet most studies have relied on “once‐only” body mass index (BMI) values, which do not reflect the development of BMI over time. ¹⁶ By examining BMI trajectories, we aimed to provide a more comprehensive assessment of how childhood growth patterns relate to BTC risk in adulthood.

MATERIALS AND METHODS

Study population

The Copenhagen School Health Records Register (CSHRR) is a uniquely comprehensive, population‐based register, which captures health examination data on children born between 1930 and 1996, who attended school in the Copenhagen municipality, both public and private schools, which encompasses nearly the entire schoolchild population as a result of Denmark’s limited use of home schooling and high participation in state‐subsidized private schools. ¹⁷ Health examinations, conducted by school physicians and nurses, provided regular measurements of height and weight. Birth weight was obtained from parental recall or infant health books, and demonstrated high validity when compared to medical birth records. ¹⁸ Because of procedural reasons, birth weight data were only available for children born between 1936 and 1996.

The initial population included all children in the CSHRR born between 1930 and 1996 (N = 406,350). Individuals without a personal identification number were excluded (n = 43,487), followed by those who emigrated, died, or were lost to follow‐up before age 18 years or before January 1, 1978 (n = 8448). Vital status was determined via the Vital Statistics Register, ¹⁹ and only individuals who were alive and residing in Denmark at age 18 years were considered eligible. Furthermore, inclusion required at least two recorded measured height and weight values between ages 6 and 15 years, which enabled the generation of BMI trajectories; those with fewer than two BMI measurements were excluded (n = 14,266). Our final analytical cohort consisted of 171,909 males and 168,248 females (Figure 1). Birth weight analyses were performed in a subset of individuals born between 1936 and 1996 for whom birth weight data were available (n = 262,508). Follow‐up started on January 1, 1978, or at age 18 years, whichever came later, and ended on the date of a diagnosis of BTC, death, emigration, or loss to follow‐up or on December 31, 2022, whichever came first.

FIGURE 1.

Flowchart of study population selection and inclusion criteria (N = 340,157). ^†BTCs identified via the Danish Cancer Registry ICD‐10 codes: C22.0 (any morphology except 8162) and C22.1 with morphology codes 8032, 8033, 8041, 8070, 8071, 8140, 8141, 8160, 8260, 8480, 8481, 8490, or 8560 (IHBDC); C24.0 and any case with morphology code 8162 (EHBDC); C23.9 (GBC); C24.1 (AVC); and C24.9 (unspecified). C24.8 (overlapping lesions) was excluded because of the small sample size and to preserve case anonymity in rare subtypes. AVC indicates ampulla of Vater cancer; BMI, body mass index; BTC, biliary tract cancer; CSHRR, Copenhagen School Health Records Register; EHBDC, extrahepatic bile duct cancer; GBC, gallbladder cancer; ICD‐10, International Classification of Diseases, Tenth Revision; IHBDC, intrahepatic bile duct cancer.

Ethics statement

This study was approved by the Danish Data Protection Agency (P‐2019‐832). Danish law does not require ethics approval and informed consent for registry‐based studies.

Exposure assessment

BMI values at ages 6–15 years were used to identify five sex‐specific BMI trajectories: below‐average, average, above‐average, overweight, and obesity (Figure S1). Each of these trajectories encapsulates a growth pattern across childhood. As previously described, ¹⁵ the trajectories were identified via latent class trajectory models with natural splines, with knot points positioned at ages 8, 10, and 12 years (approximately 25th, 50th, and 75th percentiles). A priori, the trajectories were modeled separately by sex because males and females grow differently during childhood. For each child, a posterior probability, which indicates how well a child’s BMI trajectory fits within each of the identified trajectories, was assigned for each of the five trajectories.

For comparability with other studies, we also analyzed BMI z scores at individual ages (7 and 13 years) calculated with an internal reference. We only present results based on BMI at ages 7 and 13 years, which correspond to typical school entry and exit ages, when standardized measurements are commonly recorded. Additionally, we also classified BMI via US Centers for Disease Control and Prevention (CDC) criteria at ages 7 and 13 years. The CDC classification only has two categories: normal weight, which includes underweight (<85th BMI percentile), and overweight, which includes obesity (≥85th BMI percentile). ²⁰ Height z scores were birth cohort specific to account for secular trends in growth.

Birth weight data were analyzed as a continuous variable, modeled per 500‐g increase (corresponding to approximately 1 standard deviation).

Outcome measurement

BTCs were identified via linkage with the Danish Cancer Registry with International Classification of Diseases, Tenth Revision codes available from 1978 onward. ²¹ GBC was defined by C23.9, IHBDC by C22, EHBDC by C24.0, AVC by C24.1, and BTC, unspecified by C24.9. Category C24.8, which pertains to overlapping lesions of the biliary tract, was not examined separately because of its small sample size (Table 1).

TABLE 1.

Number of cases by sex and ICD‐10 code.

Sex	ICD‐10 code	ICD description	Cases, No.
Males	C22.1 a and C22.0 a	Intrahepatic bile duct carcinoma	111
	C23.9	Malignant neoplasm of gallbladder	32
	C24.0 b	Extrahepatic bile duct carcinoma	91
	C24.1	Ampulla of Vater cancer	48
	C24.8	Overlapping lesion of biliary tract	9
	C24.9	Biliary tract cancer, unspecified	52
Total			343
Females	C22.1 a , b and C22.0 a	Intrahepatic bile duct carcinoma	108
	C23.9	Malignant neoplasm of gallbladder	69
	C24.0 b	Extrahepatic bile duct carcinoma	63
	C24.1	Ampulla of Vater cancer	42
	C24.8	Overlapping lesion of biliary tract	10
	C24.9	Biliary tract cancer, unspecified	46
Total			338
Grand total			681

Abbreviation: ICD‐10, International Classification of Diseases, Tenth Revision.

^a Intrahepatic bile duct carcinoma included C22.1 and C22.0 with morphology codes 8032, 8033, 8041, 8070, 8071, 8140, 8141, 8160, 8260, 8480, 8481, 8490, or 8560, which reflect intrahepatic bile duct origin.

^b Extrahepatic bile duct carcinoma included C24.0 and any case with morphology code 8162, regardless of topography, given that 8162 denotes perihilar (extrahepatic) origin.

Statistical analysis

For descriptive statistics, children were assigned to the BMI trajectory with the highest posterior probability (i.e., modal assignment). Medians and interquartile ranges are presented by childhood BMI trajectory and by sex. Sex‐specific Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations of the early‐life body size measures with BTC risk. In the analyses of childhood BMI trajectories, posterior probabilities were used as exposures to account for some potential misclassification. Age was the underlying timescale. Cox models were stratified by birth cohort, which allows the baseline hazard to differ by birth cohort. The proportional hazards assumption of the association between BMI trajectory and BTC risk was tested by investigating whether the associations differed across tertiles of age at risk via the likelihood ratio test. We did not detect any violations of the assumption (all p values of ≥.50).

Besides BTC overall, we estimated associations by BTC subsite for z scores and CDC categorization. In accordance with national and institutional regulations, estimates are not reported for categories containing fewer than five individuals to preserve anonymity. Therefore, some cells are not presented for BTC subsites by CDC categorization or for BMI trajectories.

RESULTS

Among 340,157 individuals (171,909 males; 168,248 females), 681 incident BTC cases were identified, including 173 IHBDC, 154 EHBDC, 101 GBC, 90 AVC, and 98 unspecified cases (Figure 1). Descriptive analyses were stratified by sex and childhood BMI trajectory (Table 2). Among males, most were classified in the average BMI trajectory (41.5%), followed by the above‐average (24.7%), below‐average (23.1%), overweight (8.5%), and obesity (2.2%) BMI trajectories. The distribution was comparable for women: average (39.0%), above‐average (26.4%), below‐average (21.2%), overweight (10.6%), and obesity (2.9%). Individuals in the overweight and obesity BMI trajectories had later median birth years, which reflects an increasing prevalence of childhood obesity in more recent cohorts. Among males, the median BMI at age 7 years ranged from 14.3 kg/m² in the below‐average trajectory to 19.3 kg/m² in the obesity trajectory, and females showed a similar pattern (from 14.1 to 19.4 kg/m²).

TABLE 2.

Descriptive characteristics of the analytical population by sex and childhood BMI trajectory.

Sex	Characteristic	Total, No.	BMI trajectory
			Below‐average	Average	Above‐average	Overweight	Obesity
Males	No. (%)	171,909	39,672 (23.1)	71,356 (41.5)	42,392 (24.7)	14,684 (8.5)	3805 (2.2)
	Year of birth, median (IQR)	171,909	1953 (1942–1970)	1950 (1941–1967)	1952 (1942–1970)	1960 (1946–1982)	1980 (1957–1990)
	BMI at age 7 years, median (IQR)	160,809	14.3 (13.9–14.7)	15.3 (14.9–15.7)	16.3 (15.7–16.8)	17.2 (16.5–18.0)	19.3 (18.2–20.7)
	BMI at age 13 years, median (IQR)	143,920	16.0 (15.4–16.4)	17.6 (17.1–18.1)	19.5 (18.9–20.2)	22.1 (21.3–23.1)	26.1 (24.9–27.6)
	Biliary tract cancer, No. (%)	343	69 (20.1)	94 (27.4)	139 (40.5)	32 (9.3)	9 (2.6)
Females	No. (%)	168,248	35,701 (21.2)	65,542 (39.0)	44,335 (26.4)	17,812 (10.6)	4858 (2.9)
	Year of birth, median (IQR)	168,248	1954 (1943–1971)	1950 (1941–1967)	1951 (1942–1968)	1956 (1944–1977)	1973 (1952–1988)
	BMI at age 7 years, median (IQR)	157,134	14.1 (13.6–14.5)	15.1 (14.6–15.5)	16.0 (15.5–16.6)	17.3 (16.5–18.1)	19.4 (18.2–20.7)
	BMI at age 13 years, median (IQR)	142,846	16.0 (15.4–16.5)	17.9 (17.3–18.5)	19.9 (19.2–20.6)	22.4 (21.6–23.4)	26.3 (25.1–27.9)
	Biliary tract cancer, No. (%)	338	55 (16.3)	136 (40.2)	102 (30.2)	32 (9.4)	13 (3.8)

Note: For descriptive purposes, individuals were assigned to the trajectory with the highest posterior probability (i.e., a measure of how well a child’s BMI trajectory fits within the identified trajectories).

Abbreviations: BMI, body mass index; IQR, interquartile range.

In males, BTC cases were distributed across childhood BMI trajectories as follows: 40.5% occurred in the above‐average trajectory, 9.3% in the overweight trajectory, and 2.6% in the obesity trajectory. In females, 30.2% of BTC cases were in the above‐average trajectory, whereas 9.5% and 3.8% were in the overweight and obesity trajectories, respectively.

In females, compared to the average BMI trajectory, the obesity trajectory was associated with an increased risk of BTCs (HR, 2.88; 95% CI, 1.62–5.15) (Figure 2). Similarly, in males, the obesity versus average BMI trajectory was associated with an increased risk (HR, 3.22; 95% CI, 1.61–6.44), and the overweight trajectory was also associated with an elevated risk of BTCs (HR, 1.58; 95% CI, 1.06–2.34). In addition, among males, there were also indications that the above‐average BMI trajectory was associated with a higher BTC risk compared to the average BMI trajectory. No significant associations were observed between BTC risk and the below‐average childhood BMI trajectory in either males or females.

FIGURE 2.

Association between childhood BMI trajectories and BTC risk in females and males. Hazard ratios for BTC risk by sex‐specific BMI trajectories are shown. Models used age as the underlying timescale, and were stratified by birth cohort. Error bars represent 95% CIs. BMI indicates body mass index; BTC, biliary tract cancer; CI, confidence interval.

In models using the CDC‐defined BMI status for children as the exposure, having overweight (including obesity) in childhood was associated with an increased risk of overall BTC in both sexes compared to those with normal weight (Table 3). At age 7 years, males with overweight had a more than 2‐fold increase in risk (HR, 2.05; 95% CI, 1.27–3.29), and females with overweight had a 76% higher risk (HR, 1.76; 95% CI, 1.02–3.05), compared with children with normal weight. At age 13 years, the association persisted (males: HR, 1.90; 95% CI, 1.14–3.17; females: HR, 2.20; 95% CI, 1.33–3.64).

TABLE 3.

Sex‐specific associations between BMI status in childhood based on the CDC classification (with normal weight as reference) and risk of overall BTC and BTC subsites.

BTC type	Age, years	CDC group a	Males					Females
			No.	Cases, No.	HR b	Lower 95% CI	Upper 95% CI	No.	Cases, No.	HR b	Lower 95% CI	Upper 95% CI
Overall	7	Normal weight	148,361	300	1.00	Reference		144,698	290	1.00	Reference
		Overweight	12,449	27	2.05	1.27	3.29	12,437	27	1.76	1.02	3.05
	13	Normal weight	133,730	290	1.00	Reference		131,947	286	1.00	Reference
		Overweight	10,191	30	1.90	1.14	3.17	10,904	33	2.20	1.33	3.64
Intrahepatic bile duct carcinoma	7	Normal weight	148,361	93	1.00	Reference		—	—	—	—	—
		Overweight	12,449	13	2.66	1.48	4.75	—	—	—	—	—
	13	Normal weight	133,730	95	1.00	Reference		—	—	—	—	—
		Overweight	10,191	11	2.40	1.28	4.49	—	—	—	—	—
Malignant neoplasm of gallbladder	7	Normal weight	—	—	—	—	—	144,698	61	1.00	Reference
		Overweight	—	—	—	—	—	12,437	5	1.53	0.61	3.82
	13	Normal weight	—	—	—	—	—	131,947	61	1.00	Reference
		Overweight	—	—	—	—	—	10,904	6	1.79	0.77	4.16
Extrahepatic bile duct carcinoma	7	Normal weight	148,361	81	1.00	Reference		144,698	51	1.00	Reference
		Overweight	12,449	6	1.37	0.60	3.14	12,437	10	3.83	1.94	7.56
	13	Normal weight	133,730	80	1.00	Reference		131,947	48	1.00	Reference
		Overweight	10,191	5	1.30	0.53	3.22	10,904	7	2.74	1.23	6.06
Unspecified	7	Normal weight	148,361	43	1.00	Reference		144,698	34	1.00	Reference
		Overweight	12,449	5	2.21	0.87	5.60	12,437	6	3.22	1.35	7.71
	13	Normal weight	133,730	43	1.00	Reference		131,947	39	1.00	Reference
		Overweight	10,191	7	3.52	1.57	7.87	10,904	5	2.22	0.87	5.66

Note: In accordance with national and institutional regulations, estimates are not reported for categories containing fewer than five individuals to preserve anonymity (represented by the dashes). For several subsites (e.g., ampulla of Vater and overlapping lesion of the biliary tract), these could not be shown for either sex, for the same reason.

Abbreviations: BMI, body mass index; BTC, biliary tract cancer; CDC, US Centers for Disease Control and Prevention; CI, confidence interval; HR, hazard ratio.

^a Normal weight includes both underweight and normal weight, and overweight includes both overweight and obesity, on the basis of CDC BMI classification criteria.

^b Age was the underlying timescale, and models were Cox stratified by birth cohort.

In our examination of individual BTC subsites, we consistently observed positive associations with childhood overweight versus normal weight (Table 3). For example, among males, having overweight at age 7 years (HR, 2.66; 95% CI, 1.48–4.75) and age 13 years (HR, 2.40; 95% CI, 1.28–4.49) was associated with an increased risk of IHBDC. Among females, the strongest association was observed for EHBDC among those with overweight at age 7 years (HR, 3.83; 95% CI, 1.94–7.56), with an elevated risk persisting among those with overweight at age 13 years (HR, 2.74; 95% CI, 1.23–6.06) (Table 3). Additionally, we observed strong associations for unspecified BTC, with HR estimates of 3.52 (95% CI, 1.57–7.87) for males who had overweight at age 13 years and 3.22 (95% CI, 1.35–7.71) for females who had overweight at age 7 years. The HRs for GBC in females and EHBDC in males had positive associations, although they were limited by small sample sizes. In accordance with national and institutional regulations, estimates are not reported for categories containing fewer than five individuals to preserve anonymity, and as a result several subsites, including AVC and overlapping lesions of the biliary tract, could not be shown for either sex.

The patterns observed with childhood BMI z scores as continuous variables were similar to the BMI categorization analyses with positive associations per one‐unit increase in z score (Table S1). Linear growth, assessed with childhood height z scores as continuous variables, was also examined for its associations with BTC risk, which allowed us to present estimates for all subsites because the continuous variables provide more power (Table 4). Among males, each one‐unit increase in height z score at age 13 years was associated with a 16% higher risk of BTCs (HR, 1.16; 95% CI, 1.04–1.29). The strongest associations were seen for GBC (HR, 1.53; 95% CI, 1.05–2.23; at age 13 years) and BTCs of unspecified location (HR, 1.45; 95% CI, 1.09–1.92; at age 13 years). In contrast, HRs tended to be around the null for associations between childhood height and BTC risk in females. Birth weight was not associated with BTC risk in either males or females (Table S2).

TABLE 4.

Sex‐specific associations between childhood height z scores (per one‐unit increase) and risk of overall BTC and BTC subsites.

BTC type	Age, years	Males					Females
		No.	Cases, No.	HR a	Lower 95% CI	Upper 95% CI	No.	Cases, No.	HR a	Lower 95% CI	Upper 95% CI
Overall	7	160,809	327	1.16	1.04	1.29	157,134	317	1.04	0.93	1.17
	13	143,920	320	1.16	1.04	1.30	142,846	319	1.05	0.94	1.17
Intrahepatic bile duct carcinoma	7	160,809	106	1.02	0.84	1.23	157,134	104	0.99	0.81	1.20
	13	143,920	106	1.01	0.83	1.22	142,846	105	1.08	0.89	1.31
Malignant neoplasm of gallbladder	7	160,809	31	1.37	0.96	1.96	157,134	66	0.97	0.76	1.24
	13	143,920	28	1.53	1.05	2.23	142,846	67	0.95	0.75	1.22
Extrahepatic bile duct carcinoma	7	160,809	87	1.16	0.94	1.44	157,134	61	1.14	0.89	1.47
	13	143,920	85	1.15	0.93	1.43	142,846	55	1.11	0.85	1.45
Ampulla of Vater cancer	7	160,809	46	1.19	0.88	1.59	157,134	37	0.96	0.70	1.33
	13	143,920	42	1.18	0.87	1.61	142,846	40	0.87	0.64	1.19
Unspecified	7	160,809	48	1.43	1.07	1.90	157,134	40	1.11	0.81	1.52
	13	143,920	50	1.45	1.09	1.92	142,846	44	1.24	0.92	1.68

Abbreviations: BMI, body mass index; BTC, biliary tract cancer; CI, confidence interval; HR, hazard ratio.

^a Age was the underlying timescale, and models were Cox stratified by birth cohort.

DISCUSSION

Our research emphasizes the associations of childhood BMI trajectories, particularly overweight and obesity trajectories, with an increased risk of adult BTCs. We also observed an increased risk of BTCs associated with childhood tallness among males. Our childhood BMI findings extend and build upon previous literature that has examined adult obesity and BTC risk. ³ , ²² Several large cohort and pooled analyses have shown that adult adiposity is associated with increased BTC risk, ³ , ¹⁰ , ²² , ²³ similar to what we found for childhood obesity. However, to our knowledge, no prior epidemiologic studies have specifically investigated childhood BMI trajectories as a risk factor for BTCs. Although BTC‐specific investigations are limited, our findings are consistent with a growing body of evidence linking childhood obesity to long‐term cancer risk. ¹³ , ¹⁶ , ²⁴ Numerous large‐scale cohort studies and pooled analyses have consistently demonstrated a positive association between higher childhood BMI and cancer risk. ¹³ , ¹⁵ , ²⁴ , ²⁵ Our findings expand upon this literature by identifying BTCs as cancers affected by early‐life adiposity. These findings underscore the importance of monitoring and intervening in early‐life BMI to reduce long‐term BTC risk.

The biological plausibility of our findings is well supported by established mechanisms linking obesity with cancer. Obesity induces metabolic disturbances, including chronic inflammation, insulin resistance, dyslipidemia, and altered bile acid metabolism, that contribute to hepatobiliary carcinogenesis. ²⁶ , ²⁷ , ²⁸ The proinflammatory state associated with obesity is marked by elevated circulating levels of cytokines such as interleukin 6 and tumor necrosis factor α. ⁷ , ⁸ , ¹⁴ , ²⁹ In parallel, obesity‐induced microbial dysbiosis has emerged as a critical mediator of inflammation ¹⁴ and cancer risk. ⁷ , ²⁸ , ³⁰ Obesity‐associated microbial dysbiosis promotes systemic inflammation via increased lipopolysaccharide (LPS) translocation, which results from reduced microbial diversity and compromised gut barrier integrity. ¹⁴ , ³¹ Circulating LPS activates innate immune responses in metabolic tissues, which induce cytokine production and genotoxic stress, which contribute to a protumorigenic microenvironment. ¹⁴ , ³² Obesity‐related hyperinsulinemia and insulin‐like growth factor 1 (IGF‐1) activation drive cell proliferation and inhibit apoptosis via the PI3K‐Akt and MAPK pathways, ³³ , ³⁴ which are critical in the development of gastrointestinal malignancies. ³⁵ , ³⁶

Evidence from studies of metabolic dysfunction and cholangiocarcinogenesis supports the relevance of metabolic pathways in the development of BTCs. ³⁰ , ³⁷ , ³⁸ Alterations in lipid metabolism, insulin resistance, chronic inflammation, and disrupted energy regulation have been identified as important contributors to carcinogenic processes in adult populations. ¹³ , ¹⁵ , ²⁵ Childhood obesity may heighten lifelong cancer risk by extending exposure to these metabolic disturbances. ¹³ , ¹⁵ , ²⁵ When considered alongside the present findings, these data suggest that metabolic vulnerability may begin early in life, and may continue to influence cancer risk across the life course. Future research incorporating early‐life metabolic biomarkers may help clarify how adiposity‐related metabolic perturbations contribute to the development of BTCs. Childhood obesity is strongly linked to adult obesity; children with obesity are approximately five times more likely to have obesity in adulthood compared to peers without obesity. ³⁹ However, epidemiologic evidence indicates that the majority of adults with obesity did not have obesity in childhood. ³⁹ This suggests that early‐life adiposity may represent an independent etiologic factor in disease development, rather than serving solely as a surrogate for adult body size. Therefore, it is likely that not just childhood obesity but cumulative lifelong exposure could increase the risk for conditions such as BTC. This highlights the importance of interventions early in life, and supports the study of early‐life risk factors when considering strategies to reduce long‐term disease risk. ³⁹

We also observed a modest association between increased childhood height and BTC risk, particularly among males for GBC and unspecified BTC, which suggests growth‐related mechanisms. Childhood height has been associated with an increased risk of several cancer types, such as colorectal, esophageal, thyroid, breast, ovarian, prostate, and endometrial cancers. ¹³ , ²⁴ , ⁴⁰ Notably, Engeland et al. found no association between adult height and GBC among men or women in a large Norwegian cohort study. ⁴¹ In contrast, Campbell et al. found a 10% increased GBC risk per 5‐cm height gain, with a slightly stronger, nonsignificant association in men. ⁹ In our study, males consistently demonstrated positive associations between childhood height and BTCs but no associations were observed in females. Importantly, Engeland et al. and Campbell et al. focused on adult anthropometry, without addressing early‐life exposures. By building on these studies, the present study underscores the etiologic relevance of early‐life linear growth as an indicator for BTC susceptibility.

Height serves as a proxy for early‐life exposure to growth‐promoting hormones, especially IGF‐1, ⁴² which has been implicated in the development of several cancers. ¹⁴ IGF‐1 plays a central role in metabolic regulation, cellular proliferation, and apoptosis. ¹⁴ , ⁴³ Its mitogenic and antiapoptotic effects, along with genetic and epigenetic regulation, link it to both developmental and carcinogenic pathways. ¹⁴ Circulating IGF‐1 levels vary by sex, age, body composition, and hormonal milieu. ³⁵ Despite lower systemic IGF‐1 in obesity, adipose‐derived IGF‐1 may enhance paracrine signaling, and thus promote sex‐specific cancer susceptibility via tissue‐specific effects. ¹⁴ , ³⁵ , ⁴⁴ Although additional research is warranted, childhood height may serve as a proxy for sex‐specific hormonal, immune, or metabolic pathways relevant to adult BTC risk. ⁴⁴

Historically, unspecified BTCs were presumed to largely represent gallbladder tumors; however, recent shifts in BTC epidemiology indicate a rising incidence of EHBDC, particularly among males. ¹ , ³ , ⁴ , ⁵ , ⁶ Furthermore, the distribution of BTC subsites varies by sex and geography. ⁴⁵ Whereas EHBDC now surpasses GBC in prevalence in several Western populations, GBC remains predominant in regions such as India, Chile, and parts of China. ⁶ , ⁴⁶ These global and sex‐based variations reinforce the necessity of detailed histologic classification to elucidate subsite‐specific risk factors, ⁴⁶ which is emphasized by the strong associations we observed for unspecified BTCs.

A major strength of this study is its use of a large, population‐based cohort with prospectively collected and directly measured anthropometric data from mandatory and regular childhood school‐based health examinations. ¹³ The prospective design and systematic health assessments allowed for repeated and standardized measurements of height and weight across childhood, which enhance the accuracy and reliability of the exposure data and support strong internal validity. ¹³ A key strength of this study is the use of BMI trajectories based on repeated childhood measurements, which capture patterns and duration of excess weight more accurately than single‐point BMI assessments. Trajectory analysis provides a more comprehensive understanding of how ongoing or cumulative childhood adiposity influences BTC risk. This approach reduces misclassification, and enhances the ability to detect meaningful associations with long‐term cancer outcomes. Moreover, the virtually complete long‐term follow‐up, enabled by systematic, mandatory registration in the high‐quality Danish Cancer Registry, minimizes loss to follow‐up and strengthens the validity of the study’s findings within this population. Nonetheless, certain limitations warrant consideration. The study population consisted of school‐attending children of Northern European descent in Denmark. Thus, the findings may not be generalizable to globally diverse populations with different genetic backgrounds and environmental exposures or regions where BTCs are more prevalent, such as India, China, and Chile. ⁴⁶ However, Denmark's universal, publicly funded health care system minimizes socioeconomic bias in health care access. ¹³ Despite the large sample, the rarity of BTCs limited detailed subsite analyses. In addition, the long interval between childhood and the onset of BTC encompasses numerous exposures during adulthood that were not captured in the available data. Adult behavioral patterns, environmental and occupational exposures, metabolic changes, and social factors may contribute to the associations observed.

In particular, we did not have data on BMI in adulthood, which precludes the assessment of BMI trajectories across the life course. Incorporating adult BMI would shift the focus toward examining BMI growth patterns across the life course and their association with BTC risk. In contrast, our objective was to evaluate whether individuals at a higher risk of BTC can be identified solely on the basis of early‐life body size. Nonetheless, we acknowledge that unmeasured adult exposures may partially mediate the observed associations. Furthermore, children born between 1930 and 1996 grew up in markedly different socioeconomic environments, nutritional contexts, health care systems, and lifestyle conditions, which could influence both early‐life growth patterns and cancer risk later in life. Nonetheless, we accounted for changes in these factors by adjusting for birth year in our analyses. A Mendelian randomization analysis of childhood obesity found some evidence of a causal association with esophageal and possibly pancreatic cancers, ³⁸ which provides support for the hypothesis that childhood obesity might increase the risk of gastrointestinal cancers.

Additionally, unmeasured factors such as early‐life diet, environment, or genetics may confound results. Among the factors that may be pertinent in childhood, primary sclerosing cholangitis and other chronic inflammatory biliary diseases are potential confounders, although their prevalence is low in pediatric populations. Gallstones, although recognized as a risk factor for BTCs, also occur infrequently in children, and historical data indicate that the prevalence of gallstones in Danish children during the mid‐to‐late 20th century was comparatively low relative to other Western countries. ⁴⁷ Furthermore, type 2 diabetes is uncommon in children, and thus unlikely to have influenced our findings. As a result, the possibility of confounding by these factors is minimal in this context.

In conclusion, our study provides compelling evidence that childhood BMI trajectories are associated with the risk of developing BTCs. Childhood overweight was most strongly associated with IHBDC in males and EHBDC in females, and showed particularly robust associations with unspecified BTCs, which suggests that subsite‐specific investigations are of importance. Amid rising childhood obesity, our findings underscore the need for early prevention targeting excess adiposity and metabolic dysfunction. Further research should clarify mechanisms linking early growth to BTC subsites, and identify targets for high‐risk groups. Given that the average BMI among children is increasing in many countries, these findings suggest that BTC incidence may increase in these populations as well. ¹³

AUTHOR CONTRIBUTIONS

Prema S. Bhattacharjee: Writing—review and editing. Jennifer L. Baker: Conceptualization; investigation; writing—original draft; methodology; validation; visualization; supervision; funding acquisition; writing—review and editing; software; formal analysis; project administration; data curation; resources. Ruth M. Pfeiffer: Writing—review and editing. Sarah S. Jackson: Writing—review and editing. Julie Aarestrup: Conceptualization; investigation; writing—original draft; methodology; validation; visualization; writing—review and editing; software; formal analysis; project administration; data curation; resources. Jill Koshiol: Conceptualization; funding acquisition; writing—original draft; writing—review and editing; project administration; resources; supervision.

CONFLICT OF INTEREST STATEMENT

Jennifer L. Baker reports consulting for and receiving grants from Novo Nordisk A/S but has no conflicts of interest relevant to this work. The other authors declare no conflicts of interest.

Supporting information

Figure S1

Table S1

Table S2

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from Statistics Denmark. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://www.dst.dk/en/TilSalg/data‐til‐forskning with the permission of Statistics Denmark.