Association of serum folate concentrations with cancer mortality among patients with arthritis: a prospective cohort study Authors: Minglong Cai, Xiaoting Wu, Huizhi Jin, Zhenyu Li, Yujing Li, Zhu Chen Published: 2024 ABSTRACT BackgroundPatients with arthritis exhibit abnormal serum folate concentrations and have a higher incidence of cancer. The aim of this study was to investigate the association of serum folate concentrations with cancer mortality among individuals with arthritis.MethodsThis cohort study included 7,514 patients with arthritis from National Health and Nutrition Examination Survey (NHANES) 1999 to 2016 and NHANES III (1988–1994). Death outcomes were ascertained by linkage to National Death Index records through 31 December 2019. Serum folate concentrations were categorized into 4 groups, group 1 (≤ 17 nmol/L); group 2 (17-≤34 nmol/L); group 3 (34-≤51 nmol/L); and group 4 (> 51 nmol/L). Cox regression models were used to calculate hazard ratios and 95% confidence intervals for the associations between folate concentrations and the risk of mortality.ResultsDuring a median follow-up of 9.75 [interquartile range, IQR, 5.75–14.5] years, there were 2,602 all-cause deaths, 949 deaths due to cardiovascular disease, and 478 deaths due to cancer. A non-linear association was observed for folate concentrations with the risk of cancer mortality (P < 0.001) among arthritis patients. Compared with the reference group 2, the hazard ratios for cancer mortality were 1.75 (95% CI, 1.15–2.69) in group 1, 1.70 (95% CI, 1.17–2.46) in group 3, and 1.60 (95% CI, 1.13–2.25) in group 4.ConclusionsThis study shows that both low and high levels of serum folate are associated with an increased risk of cancer mortality among individuals with arthritis, indicating a potential beneficial role of maintaining moderate serum folate concentrations in decreasing the risk of cancer mortality among this population.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-024-21004-8. Background Arthritis, including rheumatoid arthritis (RA) and osteoarthritis (OA), is a persistent condition with aberrant immune signaling cascades [1, 2]. Individuals with arthritis show an elevated incidence of cancer [3, 4]. Simard et al. [5] reported that patients with RA face a 2-fold increased risk of cancer mortality compared to the general population. Moreover, a shared underlying biological basis between cancer and arthritis has been proposed [6]. However, the reasons leading to the increased risk of cancer mortality in patients with arthritis remain unclear. A possible explanation is that abnormal folate concentrations in patients with arthritis may lead to increased cancer mortality. Folate (vitamin B9) is primarily involved in one-carbon metabolism and plays a crucial role in cell growth and function [7]. Previous studies have reported associations between folate and various health conditions, including anemia [8], neural tube development [9], cognitive function or dementia [10, 11], cardiovascular disease (CVD) [12–15], diabetes [16], and tumors [17–19]. Specially, folate deficiency is common among patients with arthritis, including RA [20], and low folate concentrations have been reported to be associated with a higher risk of cancer rates [21–25]. However, the association between folate and cancer mortality risk varies among different disease populations, with contradictory findings reported in several studies [26–28]. To our knowledge, the relationship between serum folate concentrations and cancer mortality among patients with arthritis remains unclear. To address these gaps, our primary objective was to prospectively investigate the association between serum folate concentrations and cancer mortality in a nationally representative sample of American adult patients with arthritis. The secondary goal was to recommend folate supplementation in patients with arthritis by determining the optimal serum folate levels. Methods Study population The study used data from the National Health and Nutrition Examination Survey (NHANES), a periodic, cross-sectional sampling survey conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention. The NHANES represents a nationally representative sample of the noninstitutionalized US civilian population. Participants in the survey provided informations on demographic and socioeconomic characteristics, and health-related behaviors and conditions through standardized questionnaires. Physical measurements and laboratory examinations were carried out by proficient medical practitioners at mobile examination centers. This research strictly followed the reporting standards delineated in the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline for cohort studies. The NHANES study protocols received ethical approval from the Institutional Review Board of the National Center of Health Statistics. All participants in the study provided informed written consent. In this study, we included patients with arthritis aged 19 years and older, and used data from NHANES III (1984–1994) and the 9 cycles spanning from 1999 to 2016. Arthritis was defined based on self-reported physician diagnosis and included rheumatoid arthritis (RA), osteoarthritis (OA), psoriatic arthritis (PsA), and other types of arthritis. Due to the small number of PsA cases, arthritis types were ultimately classified into three groups: RA, OA, and other arthritis. A total of 10,197 individuals with arthritis were initially included. After excluding participants with prevalent cancer (n = 1,710), those with missing folate information (n = 895), and those lost to follow-up (n = 78), the final cohort of participants with arthritis (n = 7,514) was included for further analysis (Fig. 1). Fig. 1Flow chart of study participants Flow chart of study participants Measurement of serum folate and vitamin B12 concentrations Serum folate and vitamin B12 concentrations were measured using different detection methods across various cycles. A detailed description of the laboratory methods can be found on NHANES website [29]. In summary, for the detection of serum folate concentrations, affinity/high-performance liquid chromatography with electrochemical (coulometric) detection was employed in 1999–2000. In NHANES III (1988–1994) and from 2001 to 2006, the Bio-Rad Laboratories Quantaphase II radioimmunoassay kit was used. In 2007–2010, a microbiological assay was employed, and in 2011–2016, isotope-dilution high-performance liquid chromatography coupled with tandem mass spectrometry was utilized. Serum vitamin B12 levels were determined using the Bio-Rad Laboratories Quantaphase II radioimmunoassay in NHANES III and from 1999 to 2006 and using the fully automated Roche electrochemiluminescence immunoassay in 2011 through 2014. As suggested, Serum folate values from 1999 to 2006 were converted to equivalent values from 2007 to 2014 using fractional polynomial regression [15, 26, 30, 31].\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{l}\:{FOL}_{adjusted}\\={10}^{**}\left(0.0188\times\:{x}^{3}-2.7109\:\times\:{x}^{-\frac{1}{2}}\:+3.8276\right),\:where\:x\:=\:{log}_{10}FOL\end{array}$$\end{document} Vitamin B12 values from NHANES III (1991–1994) and from 1999 to 2006 were converted to equivalent values from 2011 to 2014 using the Deming regression model [15, 26, 32].\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{l}\:Roche\:VitB12\\={10}^{\text{*}\text{*}}\left(0.97\times\:\text{l}\text{o}\text{g}10\left(y\right)+0.14\text{p}\text{g}/\text{m}\text{l}\right),\:where\:y\:=\:BioRad\:Vit\:B12\:\end{array}$$\end{document} Determination of mortality outcomes Death outcomes were ascertained by linkage to National Death Index records through 31 December 2019. Cause of death was determined by physicians reviewing medical records and death certificates. ICD-10 (International Statistical Classification of Diseases, 10th Revision) [33] were used to classify mortality. All-cause mortality was defined as death of any reason, CVD mortality was defined as death from CVD (I00 to I09, I11, I13, I20 to I51, and I60 to I69) and cancer mortality was defined as death from cancer (C00 to C97). Additionally, the cause of death was defined as death from the primary cause, with each patient assigned only one main cause of death. Assessment of covariates Potential confounding factors such as age, sex, ethnicity, educational level, family income, body mass index (BMI), smoking status, alcohol consumption, physical activity, and disease status were considered as covariates. Ethnicity was categorized as non-Hispanic White, Hispanic American, non-Hispanic Black, and other ethnicity. Educational level was classified as less than high school, high school or equivalent, and college or above. Family income, based on the family poverty income ratio (PIR), was categorized as poor (PIR < 1), near poor (1 ≤ PIR < 2), middle income (2 ≤ PIR < 4), and high income (PIR ≥ 4). BMI was calculated as weight in kilograms divided by height in meters squared. Smoking status was categorized as non-smoker (smoked ≤ 100 cigarettes), former smoker (smoked>100 cigarettes but currently not smoking), and current smoker (smoked > 100 cigarettes and currently smoking). Based on self-reported daily alcohol consumption, the participants were classified as rare, low to moderate, and heavy drinkers. Low to moderate drinking was defined as consuming < 2 drinks/day for men and < 1 drink/day for women. Heavy drinking was defined as consuming ≥ 2 drinks per day for men and ≥ 1 drink/day for women. Information on a physician-diagnosed history of hypertension, diabetes, and the type of arthritis was self-reported. Regarding laboratory examination data, plasma glycated hemoglobin (HbA1c), serum total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), serum cotinine, and homocysteine were measured at baseline. Additionally, information on dietary supplements taken during the past 30 days was collected by trained professionals through matching the products provided by the participants with the drug and dietary supplement database. Statistical analysis The analysis of the study data adhered to CDC guidelines [29], and sample weights, clustering, and stratification were incorporated in all analyses because of the complex sampling design of the NHANES. Serum folate concentrations were categorized into 4 groups: group 1 (≤ 17 nmol/L), group 2 (17-≤34 nmol/L), group 3 (34-≤51 nmol/L), and group 4 (> 51 nmol/L). The differences between the 4 serum folate groups were assessed using the chi-square test for categorical variables and t/Wilcoxon test for continuous variables. Multiple imputation was used because complete-case analysis may introduce selection bias [34]. Missing data for categorical variables were treated as a separate category (“missing”) in the analysis, while continuous variables were imputed with the predictive mean matching method using the mice package under the missing-at-random assumption (Supplementary Fig. 1) [35]. A total of m = 20 imputed datasets were generated, and density and convergence plots were used to diagnose the quality of the imputed values (Supplementary Fig. 2 and Supplementary Fig. 3). To account for both survival time and outcomes, Cox regression models were applied to each imputed dataset to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between serum folate concentrations and the risks of all-cause, CVD and cancer mortality. The start time in the Cox regression model was defined as the time of patient enrollment. The follow-up period ended in 2019. The primary endpoints were CVD mortality, cancer mortality, and all-cause mortality. For patients lost to follow-up, those who died from causes outside the study, or those still alive at the end of the observation period, data were recorded as censored. Subsequently, the estimates from each imputed dataset were pooled into a single set of HRs and 95%CIs. To adjust for confounding factors, four Cox models were used: Model 1 (unadjusted); Model 2, adjusted for age, sex, ethnicity, BMI, educational level, PIR, smoking status, alcohol consumption, and physical activity; Model 3, further adjusted for arthritis type, hypertension, and diabetes; and Model 4, further adjusted for HbA1c, cotinine, triglycerides, TC, HDL, and LDL. Restricted cubic spline analysis with 4 knots (at the 5th, 35th, 65th, and 95th percentiles) was employed to examine the non-linear association between serum folate concentrations and cancer mortality. The likelihood ratio test was then used to assess the significance of this relationship. We further conducted stratified analyses on the basis of the following factors: age (≤ 60 years or > 60 years); sex (Female or Male); ethnicity (non-Hispanic White or other ethnicity); BMI (≥ 30 or < 30 kg/m2); PIR (< 2 or ≥ 2); smoking status (never or ever/current); arthritis type (RA, OR or other type of arthritis); Hypertension (yes or no); Diabetes (yes or no); HbA1c (> 5.6% or ≤ 5.6%. Interaction terms between folate concentrations and the stratified factors were included in the models. We also conducted a series of sensitivity analyses: (1) Further adjusted for vitamin B12 and homocysteine; (2) Total vitamin B12 and folate intake were additionally adjusted since the folic acid intake was positively correlated with serum folate concentrations [36]; (3) Further adjusted for the inflammation marker CRP; (4) Further adjusted for estimated glomerular filtration rate (GFR), an indicator of kidney function; (5) Further adjusted for liver function markers, including levels of aspartate aminotransferase (AST), alanine transaminase (ALT), gamma-glutamyl transpeptidase (GGT), and lactate dehydrogenase (LDH); and (6) repeated the main analysis in different serum folate concentrations categories. These categories were as follows: category1 (≤ 16 nmol/L, 16-≤32 nmol/L, 32-≤48 nmol/L, and > 48 nmol/L), category2 (≤ 18, 18-≤36 nmol/L, 36-≤54 nmol/L, and > 54 nmol/L), and category3 (≤ 20 nmol/L, 20-≤40 nmol/L, 40-≤60 nmol/L, and > 60 nmol/L). All analyses were conducted in R (version 4.3.0) and a 2-sided P < 0.05 was set as the threshold for statistical significance. The data analysis was performed between April 1, 2023, and December 5, 2023. Results Baseline characteristics of the study participants In the cohort of 7,514 patients with arthritis with a median follow-up of 9.75 [interquartile range (IQR), 5.75–14.5] years, there were 2,602 all-cause deaths, 949 CVD deaths, and 478 cancer-related deaths. The patients mean age was 58.25 ± 14.32 years, with males constituting 37.06% of the cohort. The median [IQR] serum folate concentration was 39.41 [25.50-59.95] nmol/L. Baseline characteristics of the study population according to serum folate status were shown in Table 1. Male patients, non-Hispanic Black individuals, those with a higher BMI, lower educational and economic level, smokers, high alcohol consumption and patients with diabetes had significantly lower serum folate concentrations. Among laboratory indicators, patients with lower folate concentrations were more likely to have elevated TC, LDL, and lower HDL levels. Among different type of arthritis, patients with RA had the lowest serum folate concentrations. Based on the World Health Organization (WHO) definition of folate deficiency (serum folate levels below 10 nmol/L) [42], our analysis revealed that 179/2597 (6.89%) of RA patients exhibited folate deficiency compared to 122/3869 (3.15%) of OA patients. Moreover, serum folate concentrations were inversely correlated with homocysteine concentrations (R = -0.05, P = 0.009). Table 1Baseline characteristics of patients with arthritis by serum folate concentrations in NHANES III (1988–1994) and NHANES 1999 through 2016Serum folate level (nmol/L)CharacteristicsTotalGroup 1(≤ 17)Group 2(17-≤34)Group 3(34-≤51)Group 4(> 51) P Age, Mean (SD), year58.25 (14.32)53.60 (13.89)54.93 (14.42)56.32 (14.27)63.24 (12.89)< 0.001 Sex (%) < 0.001 male2810(37.40)343 (39.24)900 (41.65)734 (38.63)833 (32.30) female4704 (62.60)531 (60.76)1261 (58.35)1166 (61.37)1746 (67.70) Ethnicity (%) < 0.001 No-Hispanic White4101 (54.6)398 (45.54)1037 (47.99)985 (51.84)1681 (65.18) Hispanic American1009 (13.4)129 (14.76)304 (14.07)286 (15.05)290 (11.24) Others835 (11.1)43 (4.92)276 (12.77)237 (12.47)279 (10.82) No-Hispanic Black1569 (20.9)304 (34.78)544 (25.17)392 (20.63)329 (12.76) Education (%) < 0.001 Less than high school2418 (32.2)340 (38.90)777 (35.96)623 (32.79)678 (26.29) High school or equivalent1862 (24.8)284 (32.49)864 (39.98)830 (43.68)1245 (48.27) College or above3223 (42.9)246 (28.15)516 (23.88)446 (23.47)654 (25.36) missing11 (0.1)4 (0.46)4 (0.19)1 (0.05)2 (0.08) Family income to poverty ratio (%) < 0.001 poor1463 (19.5)223 (25.51)497 (23.00)360 (18.95)383 (14.85) nearPoor2004 (26.7)246 (28.15)569 (26.33)502 (26.42)687 (26.64) middle income1847 (24.6)212 (24.26)521 (24.11)444 (23.37)670 (25.98) high income1641 (21.8)118 (13.50)415 (19.20)451 (23.74)657 (25.47) missing559 (7.4)75 (8.58)159 (7.36)143 (7.53)182 (7.06) Smoking status (%) < 0.001 non-smoker3481 (46.3)339 (38.79)962 (44.52)863 (45.42)1317 (51.07) former2559 (34.1)240 (27.46)706 (32.67)663 (34.89)950 (36.84) current1339 (17.8)290 (33.18)466 (21.56)330 (17.37)253 (9.81) missing135 (1.8)5 (0.57)27 (1.25)44 (2.32)59 (2.29) Alcohol Consumption (%) < 0.001 rare2007 (26.7)103 (11.78)529 (24.48)513 (27.00)862 (33.42) low to moderate475 (6.3)187 (21.40)727 (33.64)718 (37.79)859 (33.31) heavy2491 (33.2)567 (64.87)774 (35.82)536 (28.21)664 (25.75) missing2541 (33.8)17 (1.95)131 (6.06)133 (7.00)194 (7.52) Hypertension (%) 0.001 yes4096 (54.5)439 (50.23)978 (45.26)896 (47.16)1091 (42.30) no3404 (45.3)434 (49.66)1180 (54.60)1001 (52.68)1481 (57.43) missing14 (0.2)1 (0.11)3 (0.14)3 (0.16)7 (0.27) Physical activity (%) < 0.001 active1459 (19.4)88 (10.07)418 (19.34)385 (20.26)568 (22.02) inactive2474 (32.9)224 (25.63)797 (36.88)611 (32.16)842 (32.65) missing3581 (47.7)562 (64.30)946 (43.78)904 (47.58)1169 (45.33) Diabetes (%) < 0.001 yes1452 (19.3)121 (13.84)428 (19.81)404 (21.26)499 (19.35) no5851 (77.9)737 (84.32)1683 (77.88)1443 (75.95)1988 (77.08) missing211 (2.8)16 (1.83)50 (2.31)53 (2.79)92 (3.57) Arthritis type (%) < 0.001 OA3869 (51.5)368 (42.11)1042 (48.22)965 (50.79)1494 (57.93) RA2597 (34.6)438 (50.11)791 (36.60)633 (33.32)735 (28.50) Other arthritis1048 (13.9)68 (7.78)328 (15.18)302 (15.89)350 (13.57)BMI, Mean (SD), kg/m230.28 (7.48)30.54 (7.80)31.30 (8.26)30.09 (7.15)29.59 (6.90)< 0.001TG, Median (IQR), mmol/L1.47 (1.01–2.21)1.48 (1.00-2.29)1.47 (0.99–2.26)1.45 (0.99–2.10)1.48 (1.02–2.22)0.256TC, Mean (SD), mmol/L5.26 (0.02)5.41 (1.20)5.31 (1.22)5.22 (1.08)5.23 (1.08)0.02HDL, Mean (SD), mmol/L1.40 (0.01)1.31 (0.41)1.34 (0.43)1.40 (0.44)1.45 (0.46)< 0.001LDL, Mean (SD), mmol/L3.07 (0.02)3.30 (1.02)3.11 (0.97)3.01 (0.95)3.03 (0.95)0.005Cotinine, Median (IQR), ng/ml0.06 (0.02–6.52)0.69 (0.08–212.50)0.11 (0.03–94.30)0.06 (0.02–4.52)0.04 (0.01–0.14)< 0.001HbA1c, Mean (SD), %5.76 (0.02)5.68 (1.01)5.75 (1.04)5.82 (1.09)5.76 (0.90)0.149Homocysteine, Mean (SD), umol/L9.94 (0.27)13.00 (7.36)9.86 (3.74)9.70 (4.69)9.61 (8.20)0.001Vitamin B12, Median (IQR), pg/ml544.00 (400.78-746.51)474.18 (359.98-641.29)480.32 (365.90-635.48)513.80 (381.66-693.55)654.93 (479.20-872.99)< 0.001BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HbA1c, glycated hemoglobin; TG, triglycerides; TC, total cholesterol; HDL, high-density lipoprotein; LDL, low-density lipoprotein; RA, rheumatoid arthritis; OA, osteoarthritis. SI conversion factors: to convert serum levels of vitamin B12 concentrations to picomoles per liter, multiply by 0.7378, and HbA1c levels to a proportion of total hemoglobin, multiply by 0.01 Table 1Baseline characteristics of patients with arthritis by serum folate concentrations in NHANES III (1988–1994) and NHANES 1999 through 2016Serum folate level (nmol/L)CharacteristicsTotalGroup 1(≤ 17)Group 2(17-≤34)Group 3(34-≤51)Group 4(> 51) P Age, Mean (SD), year58.25 (14.32)53.60 (13.89)54.93 (14.42)56.32 (14.27)63.24 (12.89)< 0.001 Sex (%) < 0.001 male2810(37.40)343 (39.24)900 (41.65)734 (38.63)833 (32.30) female4704 (62.60)531 (60.76)1261 (58.35)1166 (61.37)1746 (67.70) Ethnicity (%) < 0.001 No-Hispanic White4101 (54.6)398 (45.54)1037 (47.99)985 (51.84)1681 (65.18) Hispanic American1009 (13.4)129 (14.76)304 (14.07)286 (15.05)290 (11.24) Others835 (11.1)43 (4.92)276 (12.77)237 (12.47)279 (10.82) No-Hispanic Black1569 (20.9)304 (34.78)544 (25.17)392 (20.63)329 (12.76) Education (%) < 0.001 Less than high school2418 (32.2)340 (38.90)777 (35.96)623 (32.79)678 (26.29) High school or equivalent1862 (24.8)284 (32.49)864 (39.98)830 (43.68)1245 (48.27) College or above3223 (42.9)246 (28.15)516 (23.88)446 (23.47)654 (25.36) missing11 (0.1)4 (0.46)4 (0.19)1 (0.05)2 (0.08) Family income to poverty ratio (%) < 0.001 poor1463 (19.5)223 (25.51)497 (23.00)360 (18.95)383 (14.85) nearPoor2004 (26.7)246 (28.15)569 (26.33)502 (26.42)687 (26.64) middle income1847 (24.6)212 (24.26)521 (24.11)444 (23.37)670 (25.98) high income1641 (21.8)118 (13.50)415 (19.20)451 (23.74)657 (25.47) missing559 (7.4)75 (8.58)159 (7.36)143 (7.53)182 (7.06) Smoking status (%) < 0.001 non-smoker3481 (46.3)339 (38.79)962 (44.52)863 (45.42)1317 (51.07) former2559 (34.1)240 (27.46)706 (32.67)663 (34.89)950 (36.84) current1339 (17.8)290 (33.18)466 (21.56)330 (17.37)253 (9.81) missing135 (1.8)5 (0.57)27 (1.25)44 (2.32)59 (2.29) Alcohol Consumption (%) < 0.001 rare2007 (26.7)103 (11.78)529 (24.48)513 (27.00)862 (33.42) low to moderate475 (6.3)187 (21.40)727 (33.64)718 (37.79)859 (33.31) heavy2491 (33.2)567 (64.87)774 (35.82)536 (28.21)664 (25.75) missing2541 (33.8)17 (1.95)131 (6.06)133 (7.00)194 (7.52) Hypertension (%) 0.001 yes4096 (54.5)439 (50.23)978 (45.26)896 (47.16)1091 (42.30) no3404 (45.3)434 (49.66)1180 (54.60)1001 (52.68)1481 (57.43) missing14 (0.2)1 (0.11)3 (0.14)3 (0.16)7 (0.27) Physical activity (%) < 0.001 active1459 (19.4)88 (10.07)418 (19.34)385 (20.26)568 (22.02) inactive2474 (32.9)224 (25.63)797 (36.88)611 (32.16)842 (32.65) missing3581 (47.7)562 (64.30)946 (43.78)904 (47.58)1169 (45.33) Diabetes (%) < 0.001 yes1452 (19.3)121 (13.84)428 (19.81)404 (21.26)499 (19.35) no5851 (77.9)737 (84.32)1683 (77.88)1443 (75.95)1988 (77.08) missing211 (2.8)16 (1.83)50 (2.31)53 (2.79)92 (3.57) Arthritis type (%) < 0.001 OA3869 (51.5)368 (42.11)1042 (48.22)965 (50.79)1494 (57.93) RA2597 (34.6)438 (50.11)791 (36.60)633 (33.32)735 (28.50) Other arthritis1048 (13.9)68 (7.78)328 (15.18)302 (15.89)350 (13.57)BMI, Mean (SD), kg/m230.28 (7.48)30.54 (7.80)31.30 (8.26)30.09 (7.15)29.59 (6.90)< 0.001TG, Median (IQR), mmol/L1.47 (1.01–2.21)1.48 (1.00-2.29)1.47 (0.99–2.26)1.45 (0.99–2.10)1.48 (1.02–2.22)0.256TC, Mean (SD), mmol/L5.26 (0.02)5.41 (1.20)5.31 (1.22)5.22 (1.08)5.23 (1.08)0.02HDL, Mean (SD), mmol/L1.40 (0.01)1.31 (0.41)1.34 (0.43)1.40 (0.44)1.45 (0.46)< 0.001LDL, Mean (SD), mmol/L3.07 (0.02)3.30 (1.02)3.11 (0.97)3.01 (0.95)3.03 (0.95)0.005Cotinine, Median (IQR), ng/ml0.06 (0.02–6.52)0.69 (0.08–212.50)0.11 (0.03–94.30)0.06 (0.02–4.52)0.04 (0.01–0.14)< 0.001HbA1c, Mean (SD), %5.76 (0.02)5.68 (1.01)5.75 (1.04)5.82 (1.09)5.76 (0.90)0.149Homocysteine, Mean (SD), umol/L9.94 (0.27)13.00 (7.36)9.86 (3.74)9.70 (4.69)9.61 (8.20)0.001Vitamin B12, Median (IQR), pg/ml544.00 (400.78-746.51)474.18 (359.98-641.29)480.32 (365.90-635.48)513.80 (381.66-693.55)654.93 (479.20-872.99)< 0.001BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HbA1c, glycated hemoglobin; TG, triglycerides; TC, total cholesterol; HDL, high-density lipoprotein; LDL, low-density lipoprotein; RA, rheumatoid arthritis; OA, osteoarthritis. SI conversion factors: to convert serum levels of vitamin B12 concentrations to picomoles per liter, multiply by 0.7378, and HbA1c levels to a proportion of total hemoglobin, multiply by 0.01 Baseline characteristics of patients with arthritis by serum folate concentrations in NHANES III (1988–1994) and NHANES 1999 through 2016 BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HbA1c, glycated hemoglobin; TG, triglycerides; TC, total cholesterol; HDL, high-density lipoprotein; LDL, low-density lipoprotein; RA, rheumatoid arthritis; OA, osteoarthritis. SI conversion factors: to convert serum levels of vitamin B12 concentrations to picomoles per liter, multiply by 0.7378, and HbA1c levels to a proportion of total hemoglobin, multiply by 0.01 Non-linear relationships of serum folate concentrations with cancer mortality Our Cox regression model meets the assumption of proportional hazards (Supplementary Fig. 4). Besides, a non-linear relationship between serum folate concentrations and cancer mortality was observed (Fig. 2A). After adjusting for age, gender, ethnicity, BMI, educational level, PIR, smoking status, alcohol consumption, physical activity, history of hypertension and diabetes, type of arthritis, cotinine, HbA1c, TG, TC, HDL, LDL, the low folate group (group 1) had a HR of 1.75 (95% CI, 1.15–2.69) for cancer mortality, while the moderate (group 3) and high folate groups (group 4) had HRs of 1.70 (95% CI, 1.17–2.46) and 1.60 (95% CI, 1.13–2.25), respectively, compared with the reference group (group 2) (Table 2; Fig. 2B). Fig. 2Non-linear Association Between Serum Folate Concentrations and Cancer Mortality in Patients with Arthritis. (A) Restricted cubic spline analysis shows a non-linear association between serum folate concentrations and cancer mortality. We adjusted for age, sex, ethnicity, BMI, educational level, family income, smoking status, alcohol consumption, hypertension, diabetes, and type of arthritis; (B) Cumulative incidence curves show that patients with moderate level of serum folate (17-≤34 nmol/L) have a lower risk of cancer mortality. We adjusted for age, sex, ethnicity, BMI, educational level, family income, smoking status, alcohol consumption, hypertension, diabetes, and type of arthritis. Log-rank, likelihood ratio test Non-linear Association Between Serum Folate Concentrations and Cancer Mortality in Patients with Arthritis. (A) Restricted cubic spline analysis shows a non-linear association between serum folate concentrations and cancer mortality. We adjusted for age, sex, ethnicity, BMI, educational level, family income, smoking status, alcohol consumption, hypertension, diabetes, and type of arthritis; (B) Cumulative incidence curves show that patients with moderate level of serum folate (17-≤34 nmol/L) have a lower risk of cancer mortality. We adjusted for age, sex, ethnicity, BMI, educational level, family income, smoking status, alcohol consumption, hypertension, diabetes, and type of arthritis. Log-rank, likelihood ratio test Table 2Hazard ratios of Cancer, CVD, and all-cause mortality by Folate serum levels among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016Hazard ratio (95% CI)Folate serum level, nmol/L Model Group 1 (≤ 17) Group 2 (17-≤34)Group 3 (34-≤51) Group 4 (> 51) CA mortality Deaths, No./total No.96/874102/2161123/1900157/2579Model 1a1.82 (1.19–2.78)1 [reference]1.53 (1.09–2.16)1.83 (1.35–2.49)Model 2b1.77 (1.16–2.71)1 [reference]1.68 (1.17–2.41)1.59 (1.14–2.23)Model 3c1.76 (1.15–2.69)1 [reference]1.69 (1.17–2.44)1.60 (1.13–2.25)Model 4d1.76 (1.15–2.69)1 [reference]1.70 (1.17–2.47)1.60 (1.13–2.26) CVD mortality Deaths, No./total No144/ 874274/2161198/1900333/2579Model 1a0.98 (0.73–1.31)1 [reference]0.88 (0.71–1.1)1.21 (0.97–1.51)Model 2b1.05 (0.82–1.33)1 [reference]0.92 (0.72–1.19)0.82 (0.65–1.04)Model 3c1.09 (0.84–1.41)1 [reference]0.94 (0.73–1.21)0.83 (0.65–1.04)Model 4d1.09 (0.85–1.41)1 [reference]0.92 (0.72–1.19)0.82 (0.65–1.03) All-cause mortality Deaths, No./total No383/874686/2161581/1900952/2579Model 1a0.96 (0.8–1.16)1 [reference]1.06 (0.92–1.23)1.49 (1.3–1.71)Model 2b1.03 (0.88–1.20)1 [reference]1.10 (0.94–1.28)1.05 (0.91–1.20)Model 3c1.06 (0.9–1.25)1 [reference]1.11 (0.95–1.29)1.05 (0.91–1.21)Model 4d1.07 (0.91–1.25)1 [reference]1.09 (0.94–1.27)1.03 (0.9–1.19)a Unadjusted; b Adjusted for age (continuous), sex (male or female), ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic American, or other), body mass index (calculated as weight in kilograms divided by height in meters squared, continuous), educational level (less than high school, high school or equivalent, or college or above), family income level (poor, near poor, middle income, or high income), smoking status (never, ever, or current), alcohol consumption (rare, low to moderate, or heavy), physical activity (active or inactive); c Further adjusted for hypertension (yes or no), diabetes (yes or no), arthritis type (RA, OA or other arthritis); d Further adjusted for HbA1c (continuous), serum cotinine (continuous), homocysteine (continuous), TG (continuous), TC (continuous), HDL (continuous), and LDL (continuous). CVD, cardiovascular disease; NHANES, National Health and Nutrition Examination Survey Table 2Hazard ratios of Cancer, CVD, and all-cause mortality by Folate serum levels among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016Hazard ratio (95% CI)Folate serum level, nmol/L Model Group 1 (≤ 17) Group 2 (17-≤34)Group 3 (34-≤51) Group 4 (> 51) CA mortality Deaths, No./total No.96/874102/2161123/1900157/2579Model 1a1.82 (1.19–2.78)1 [reference]1.53 (1.09–2.16)1.83 (1.35–2.49)Model 2b1.77 (1.16–2.71)1 [reference]1.68 (1.17–2.41)1.59 (1.14–2.23)Model 3c1.76 (1.15–2.69)1 [reference]1.69 (1.17–2.44)1.60 (1.13–2.25)Model 4d1.76 (1.15–2.69)1 [reference]1.70 (1.17–2.47)1.60 (1.13–2.26) CVD mortality Deaths, No./total No144/ 874274/2161198/1900333/2579Model 1a0.98 (0.73–1.31)1 [reference]0.88 (0.71–1.1)1.21 (0.97–1.51)Model 2b1.05 (0.82–1.33)1 [reference]0.92 (0.72–1.19)0.82 (0.65–1.04)Model 3c1.09 (0.84–1.41)1 [reference]0.94 (0.73–1.21)0.83 (0.65–1.04)Model 4d1.09 (0.85–1.41)1 [reference]0.92 (0.72–1.19)0.82 (0.65–1.03) All-cause mortality Deaths, No./total No383/874686/2161581/1900952/2579Model 1a0.96 (0.8–1.16)1 [reference]1.06 (0.92–1.23)1.49 (1.3–1.71)Model 2b1.03 (0.88–1.20)1 [reference]1.10 (0.94–1.28)1.05 (0.91–1.20)Model 3c1.06 (0.9–1.25)1 [reference]1.11 (0.95–1.29)1.05 (0.91–1.21)Model 4d1.07 (0.91–1.25)1 [reference]1.09 (0.94–1.27)1.03 (0.9–1.19)a Unadjusted; b Adjusted for age (continuous), sex (male or female), ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic American, or other), body mass index (calculated as weight in kilograms divided by height in meters squared, continuous), educational level (less than high school, high school or equivalent, or college or above), family income level (poor, near poor, middle income, or high income), smoking status (never, ever, or current), alcohol consumption (rare, low to moderate, or heavy), physical activity (active or inactive); c Further adjusted for hypertension (yes or no), diabetes (yes or no), arthritis type (RA, OA or other arthritis); d Further adjusted for HbA1c (continuous), serum cotinine (continuous), homocysteine (continuous), TG (continuous), TC (continuous), HDL (continuous), and LDL (continuous). CVD, cardiovascular disease; NHANES, National Health and Nutrition Examination Survey Hazard ratios of Cancer, CVD, and all-cause mortality by Folate serum levels among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016 a Unadjusted; b Adjusted for age (continuous), sex (male or female), ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic American, or other), body mass index (calculated as weight in kilograms divided by height in meters squared, continuous), educational level (less than high school, high school or equivalent, or college or above), family income level (poor, near poor, middle income, or high income), smoking status (never, ever, or current), alcohol consumption (rare, low to moderate, or heavy), physical activity (active or inactive); c Further adjusted for hypertension (yes or no), diabetes (yes or no), arthritis type (RA, OA or other arthritis); d Further adjusted for HbA1c (continuous), serum cotinine (continuous), homocysteine (continuous), TG (continuous), TC (continuous), HDL (continuous), and LDL (continuous). CVD, cardiovascular disease; NHANES, National Health and Nutrition Examination Survey We also assessed the associations of folate concentrations with the risk of all-cause and cardiovascular mortality. Although we observed an elevated risk of all-cause mortality in patients with high folate concentrations (HR = 1.49, 95% CI, 1.30–1.71) in the univariate analysis, statistical differences disappeared after adjusting for multiple variables. Similarly, no observable association was found between serum folate concentrations and CVD mortality (Table 2). Stratified and sensitivity analyses We observed significant interactions between serum folate concentrations and sex (P = 0.003 for interaction) and ethnicity (P < 0.001 for interaction) (Table 3). In the subgroup of male patients, the HR for cancer mortality was 3.14 (95% CI, 1.74–5.66) in group 1, 2.40 (95% CI, 1.43–4.04) in group 3, and 3.13 (95% CI, 1.86–5.28) in group 4 compared with the reference group. While in the subgroup of female patients, there was no significant association between serum folate concentrations and cancer mortality. In the Non-Hispanic White subgroup, the HR for cancer mortality was 1.96 (95% CI, 1.19–3.24) in group1 and 1.90 (95% CI, 1.22–2.97) in group 3 compared with the reference group, and there was no significant association between serum folate concentrations and cancer mortality in other ethnic subgroups. Additionally, although the interaction did not reach statistical significance, low folate concentrations(group1) in RA showed an increased risk of cancer mortality, with an HR of 2.60 (95% CI, 1.46–4.63) compared to the reference group. Conversely, in patients with OA, high folate concentrations (group 3 and group 4) were associated with an increased risk of cancer mortality, with HRs of 1.64 (95% CI, 1.04–2.60) and 1.73 (95% CI, 1.12–2.66), respectively. Table 3Associations between serum folate concentrations and Cancer Mortality in various subgroups among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016Hazard ratio (95% CIs) aP value for interactionCharacteristicGroup 1(≤ 17)Group 2(17-≤34)Group 3(34-≤51)Group 4(> 51)Age, y0.389 ≤ 601.72 (0.82–3.59)1 [reference)1.72 (0.83–3.58)2.65 (1.28–5.49) >601.51 (0.93–2.46)1 [reference)1.70 (1.07–2.68)1.43 (1.03-2.00)Sex0.003 Female1.16 (0.69–1.93)1 [reference)1.29 (0.81–2.05)0.92 (0.6–1.41) Male3.14 (1.74–5.66)1 [reference)2.40 (1.43–4.04)3.13 (1.86–5.28)Ethnicity< 0.001 Non-Hispanic White1.96 (1.19–3.24)1 [reference)1.90 (1.22–2.97)1 (0.98–1.03) Other ethnicity1.47 (0.88–2.45)1 [reference)1.15 (0.69–1.91)1.4 (0.87–2.25)BMI0.09 < 301.99 (1.25–3.16)1 [reference)2.29 (1.45–3.64)1.72 (1.11–2.67) ≥ 301.06 (0.55–2.02)1 [reference)1.18 (0.68–2.05)1.43 (0.88–2.31)Family income to poverty ratio (PIR)0.143 < 22.18 (1.26–3.79)1 [reference)2.62 (1.57–4.38)1.91 (1.21-3.00) ≥ 21.57 (0.72–3.41)1 [reference)1.49 (0.83–2.66)1.46 (0.74–2.88)Smoking status0.09 Never1.40 (0.72–2.73)1 [reference)1.31 (0.71–2.41)1.15 (0.67–1.98) Ever or current1.80 (1.10–2.92)1 [reference)1.73 (1.09–2.74)1.68 (1.13–2.51)Arthritis type0.194 RA2.60 (1.46–4.63)1 [reference)1.96 (0.98–3.92)1.33 (0.74–2.37) OA1.27 (0.72–2.26)1 [reference)1.64 (1.04–2.60)1.73 (1.12–2.66) Other arthritis0.49 (0.1–2.44)1 [reference)1.07 (0.42–2.76)1.00 (0.43–2.3)Hypertension0.546 yes1.45 (0.78–2.7)1 [reference)1.82 (1.13–2.94)1.53 (0.97–2.43) no2.09 (1.11–3.92)1 [reference)1.41 (0.77–2.57)1.52 (0.84–2.76)Diabetes0.048 yes0.83 (0.3–2.27)1 [reference)1.27 (0.61–2.65)0.74 (0.33–1.68) no1.94 (1.21–3.09)1 [reference)1.82 (1.22–2.71)1.86 (1.26–2.73)HbA1c, %0.845 > 5.61.98 (1.12–3.5)1 [reference)1.33 (0.79–2.24)1.29 (0.73–2.31) ≤ 5.61.74 (1.01-3.00)1 [reference)2.04 (1.19–3.49)1.87 (1.18–2.97)a Adjusted for age, sex, ethnicity, body mass index, educational level, family income level, smoke status, alcohol consumption, physical activity, hypertension, diabetes, arthritis type, HbA1c, cotinine, homocysteine, TG, TC, HDL, and LDL. The strata variable was not included when stratifying by itself. BMI, body mass index; HbA1c, glycated hemoglobin; TC, total cholesterol, HDL, high-density lipoprotein; LDL: low-density lipoprotein Table 3Associations between serum folate concentrations and Cancer Mortality in various subgroups among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016Hazard ratio (95% CIs) aP value for interactionCharacteristicGroup 1(≤ 17)Group 2(17-≤34)Group 3(34-≤51)Group 4(> 51)Age, y0.389 ≤ 601.72 (0.82–3.59)1 [reference)1.72 (0.83–3.58)2.65 (1.28–5.49) >601.51 (0.93–2.46)1 [reference)1.70 (1.07–2.68)1.43 (1.03-2.00)Sex0.003 Female1.16 (0.69–1.93)1 [reference)1.29 (0.81–2.05)0.92 (0.6–1.41) Male3.14 (1.74–5.66)1 [reference)2.40 (1.43–4.04)3.13 (1.86–5.28)Ethnicity< 0.001 Non-Hispanic White1.96 (1.19–3.24)1 [reference)1.90 (1.22–2.97)1 (0.98–1.03) Other ethnicity1.47 (0.88–2.45)1 [reference)1.15 (0.69–1.91)1.4 (0.87–2.25)BMI0.09 < 301.99 (1.25–3.16)1 [reference)2.29 (1.45–3.64)1.72 (1.11–2.67) ≥ 301.06 (0.55–2.02)1 [reference)1.18 (0.68–2.05)1.43 (0.88–2.31)Family income to poverty ratio (PIR)0.143 < 22.18 (1.26–3.79)1 [reference)2.62 (1.57–4.38)1.91 (1.21-3.00) ≥ 21.57 (0.72–3.41)1 [reference)1.49 (0.83–2.66)1.46 (0.74–2.88)Smoking status0.09 Never1.40 (0.72–2.73)1 [reference)1.31 (0.71–2.41)1.15 (0.67–1.98) Ever or current1.80 (1.10–2.92)1 [reference)1.73 (1.09–2.74)1.68 (1.13–2.51)Arthritis type0.194 RA2.60 (1.46–4.63)1 [reference)1.96 (0.98–3.92)1.33 (0.74–2.37) OA1.27 (0.72–2.26)1 [reference)1.64 (1.04–2.60)1.73 (1.12–2.66) Other arthritis0.49 (0.1–2.44)1 [reference)1.07 (0.42–2.76)1.00 (0.43–2.3)Hypertension0.546 yes1.45 (0.78–2.7)1 [reference)1.82 (1.13–2.94)1.53 (0.97–2.43) no2.09 (1.11–3.92)1 [reference)1.41 (0.77–2.57)1.52 (0.84–2.76)Diabetes0.048 yes0.83 (0.3–2.27)1 [reference)1.27 (0.61–2.65)0.74 (0.33–1.68) no1.94 (1.21–3.09)1 [reference)1.82 (1.22–2.71)1.86 (1.26–2.73)HbA1c, %0.845 > 5.61.98 (1.12–3.5)1 [reference)1.33 (0.79–2.24)1.29 (0.73–2.31) ≤ 5.61.74 (1.01-3.00)1 [reference)2.04 (1.19–3.49)1.87 (1.18–2.97)a Adjusted for age, sex, ethnicity, body mass index, educational level, family income level, smoke status, alcohol consumption, physical activity, hypertension, diabetes, arthritis type, HbA1c, cotinine, homocysteine, TG, TC, HDL, and LDL. The strata variable was not included when stratifying by itself. BMI, body mass index; HbA1c, glycated hemoglobin; TC, total cholesterol, HDL, high-density lipoprotein; LDL: low-density lipoprotein Associations between serum folate concentrations and Cancer Mortality in various subgroups among individuals with arthritis in NHANES III (1988–1994) and NHANES 1999 through 2016 a Adjusted for age, sex, ethnicity, body mass index, educational level, family income level, smoke status, alcohol consumption, physical activity, hypertension, diabetes, arthritis type, HbA1c, cotinine, homocysteine, TG, TC, HDL, and LDL. The strata variable was not included when stratifying by itself. BMI, body mass index; HbA1c, glycated hemoglobin; TC, total cholesterol, HDL, high-density lipoprotein; LDL: low-density lipoprotein The findings remained robust in sensitivity analyses when further adjusted for homocysteine and vitamin B12, folate and vitamin B12 dietary intake and supplement, CRP, GFR, and liver function indicators (Supplementary Table 3). Moreover, the associations did not materially change when serum folate concentrations were classified into different categories (Supplementary Table 4). Discussion To the best of our knowledge, this is the first cohort study with a large sample size to investigate the association between serum folate concentrations and cancer mortality in patients with arthritis. Our study showed that both low and high serum folate concentrations were associated with an increased risk of cancer mortality. These results have potential implications for clinical practice and dietary recommendations aimed at mitigating cancer mortality in individuals with arthritis. As expected, heavy alcohol consumption was a factor which decreased substantially folate concentrations [37]. This may be due to chronic alcohol exposure impairing folate absorption by inhibiting the expression of the reduced folate carrier and reducing hepatic uptake and renal conservation of circulating folate [38]. Our finding of a notable inverse correlation between folate and homocysteine concentrations is consistent with previous reports [39, 40]. Individuals who identified as non-Hispanic Black, with a lower educational and economic status, showed reduced folate concentrations, suggesting potential nutritional deficiencies or unhealthy lifestyle habits in these demographic groups. Interestingly, in line with prior studies, low folate concentrations were negatively correlated with BMI [41, 42] and lipid levels [43], which can be explained by a previous finding that folate deficiency contributed to lipid accumulation [44]. In different arthritis populations, individuals with RA show markedly lower folate concentrations than those with OA. This difference between patients with RA and those with OA could be attributed to the use of folic acid antagonists, such as methotrexate [20]. The association between folate levels and cancer mortality has been investigated in diverse populations, but conflicting results have been observed. In the general population, Peng et al. [26] based on NHANES data from 1999 to 2010 found that low folate concentrations increased cancer mortality. Consistent with this, Yang et al. [27] used data from NHANES III (1988–1994), observed that low folate concentrations were associated with a higher cancer mortality. However, a randomized double-blind cohort study showed that supplementing folate and vitamin B12 in patients with ischemic heart disease elevated the risk of cancer mortality [28]. Therefore, the effect of folate on cancer mortality may manifest a dualistic nature, and its effect may vary across different populations. In our study, we found both low and high serum folate concentrations were associated with an increased risk of cancer mortality in patients with arthritis. Interestingly, although the interaction between folate concentrations and arthritis type did not reach statistical significance, stratified analyses revealed heterogeneity in the effect of folate concentrations on cancer mortality across arthritis subgroups. In patients with RA, low folate concentrations were associated with a higher risk of cancer mortality compared with the reference group. In contrast, high folate concentrations in OA patients were associated with an increased risk of cancer mortality. These findings suggested that the elevated cancer mortality in RA and OA may due to different folate concentrations, as the folate concentrations in RA patients is indeed lower than that in OA patients. Currently, the mechanisms underlying the association between folate concentrations and cancer mortality remain unclear. Previous research has suggested that elevated folate concentrations may promote cancer development through the methionine metabolism pathway [17]. On the other hand, low folate levels may indicate a poor nutritional status in patients, thereby affecting their prognosis. Prior studies have also reported the association of folate concentrations with all-cause mortality and CVD mortality in different populations [15, 26, 30, 31]. In our study, folate concentrations in patients with arthritis were not significantly associated with all-cause and CVD mortality. However, the stratified analyses showed a consistent finding with a previous study that low folate concentrations in RA were associated with an increased risk of all-cause and CVD mortality (Supplementary Table 1, Supplementary Table 2) [30]. A notable strength of this study is its large cohort of 7,514 arthritis patients, a long median follow-up period of 9.75 years, and the rigorous methodology using Cox regression models adjusted for multiple covariates. It uniquely identifies a non-linear association between serum folate levels and cancer mortality, addressing a critical gap in understanding the interplay between nutrition, arthritis, and oncology. The finding highlighting the potential benefit of maintaining moderate serum folate concentrations to reduce cancer mortality risk in arthritis patients. However, our study has several limitations. First, while we identified the folate concentration range in group 2 (17-≤34 nmol/L) as the reference group as having a lower cancer mortality compared with the other groups, we were unable to precisely determine the optimal folate concentration range for the lowest cancer mortality risk. Determining this range may require a larger sample size and more advanced algorithms in future studies; Second, we used serum folate concentrations, which may not accurately reflect the long-term folate status; Third, the Cox regression in this study only included baseline variables of patients, and these variables may have changed during the follow-up, potentially affecting the results; Fourth, adjusting for all potential or unknown confounding factors was challenging, even though we adjusted for a considerable number of covariates. These limitations highlight areas for improvement in future research endeavors. Conclusions This study shows a non-linear association between serum folate concentrations and cancer mortality, and that arthritis patients with both low and high levels of serum folate have a higher risk of cancer mortality. Our results indicate a potential beneficial role of maintaining moderate serum folate concentrations for decreasing the risk of cancer mortality in patients with arthritis. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1