Association between F₂-isoprostane metabolites and weight change in older women: a longitudinal analysis

Abstract

Introduction:

Theoretically, some metabolic traits may predispose older individuals to weight loss during aging, leading to increased all-cause mortality and many serious health issues. Biomarkers to robustly predict progressive weight loss during aging are, however, lacking. We prospectively assessed if urinary levels of F₂-isoprostanes and their peroxisomal β-oxidation metabolite, 2,3-dinor-5,6-dihydro-15-F_2t-isoprostane (F₂-IsoP-M), were associated with subsequent weight loss in middle-aged and older women.

Methods:

Included in the analysis were 2066 women aged 40–70 years, a subset of a prospective cohort study. F₂-isoprostanes (F₂-IsoPs) and its β-oxidation metabolite F₂-IsoP-M were measured in urine using gas chromatography-mass spectrometry. Measurements of anthropometry and exposures to major determinants of body weight were performed at baseline and repeated thrice over 15-year follow-up. The longitudinal associations of F₂-IsoP-M and the F₂-IsoP-M to its parent compound F₂-IsoP ratio (MPR) with repeatedly measured weight changes were examined using linear mixed-effect models.

Results:

After adjusting for time-varying covariates energy intake, physical activity, and comorbidity index, among others, levels of F₂-IsoP-M and the MPR were both inversely associated with percentage of weight change. Weight in the highest quartile of these two biomarkers was 1.33% (95% CI = −2.41, −0.24) and 1.09% (95% CI = −2.16, −0.02) lower than those in the lowest quartile group, with P for trend of 0.01 and 0.03, respectively. The inverse association was consistently seen across follow-up periods, although appearing stronger with prolonged follow-up. There was no association between the parent compound F₂-IsoPs and weight change.

Conclusion:

This study demonstrates the first piece of evidence to associate F₂-IsoP metabolism, peroxisomal β-oxidation, with weight loss in older women. Further investigations into the role of lipid peroxidation and peroxisomal β-oxidation in weight change among older individuals are warranted.

Introduction

Aging is accompanied with changes in body composition, including adipose tissue degradation and/or skeletal muscle atrophy, which predispose older individuals to weight loss [1]. Excess weight loss represents an important role in the development of sarcopenia, a common geriatric syndrome characterized by physical function impairment [2, 3]. Accumulating evidence shows that unintentional weight loss in older individuals is associated with increased all-cause mortality and many health issues such as cognitive impairment, infections and hip fracture [4–6]. Currently, biomarkers to robustly predict progressive weight loss during the course of aging are lacking. Such biomarkers, however, are of great interest and urgently needed, because they can provoke early intensification of prevention measures among individuals at high risk for continuing weight loss.

F₂-isoprostanes (F₂-IsoPs), a unique series of prostaglandin-like products formed by free radical oxidation of arachidonic acid, have been validated as sensitive and reliable biomarkers of in vivo lipid peroxidation in animal and human studies [7, 8]. Like prostaglandins, F₂-IsoPs are transformed to more polar metabolites by β-oxidation [9, 10], such as 2,3-dinor-5,6-dihydro-15-F_2t-isoprostane (F₂-IsoP-M), as exemplified in Figure 1. This β-oxidation is largely driven by peroxisomes as reduced levels of F₂-IsoP-M, but elevated F₂-IsoPs, were observed in patients with peroxisomal β-oxidation deficiency, e.g., Zellweger syndrome and adrenoleukodystrophy [9, 11, 12]. Emerging studies suggest that peroxisomal dysfunction may contribute to the pathogenesis of a range of age-related diseases, including neurodegenerative disorders, diabetes and cancer [13–15], and the aging process [16–18].

Figure 1.

F₂-IsoPs metabolism.

Conventionally, F₂-IsoPs are perceived as indicators of harmful oxidative stress [19]. However, an increasing body of evidence suggests that F₂-IsoP metabolites, such as F₂-IsoP-M, may reflect intensity of oxidative metabolism [20, 21]. F₂-IsoPs have been associated with excessive body fat accumulation in previous cross-sectional studies [22, 23], but F₂-IsoP-M appears to better characterize the relationship than unmetabolized F₂-IsoPs [24, 25], and has been identified as a good biomarker of peroxisomal β-oxidation [11, 26]. Ma et al. recently demonstrated a strong and inverse association between F₂-IsoP-M concentrations and body composition measures such as trunk fat mass. More importantly, they found that higher concentrations of urinary F₂-IsoPs and F₂-IsoP-M were significantly associated with lower lean body mass, indicating a link of the biomarkers with weight loss and sarcopenia development [27]. Despite the compelling cross-sectional evidence, prospective epidemiological data on the relationship between F₂-IsoPs and weight change have, however, been very limited. Only two studies evaluated and found an inverse association between F₂-IsoPs and weight change [28, 29]. These studies were conducted in Western populations, with a relatively small sample size, and measured subsequent anthropometric change at a single time point [28, 29]. Neither one specifically examined F₂-IsoP-M.

In this study, we measured urinary concentrations of F₂-IsoPs and F₂-IsoP-M in 2066 middle-aged and older women, a subset of the Shanghai Women’s Health Study (SWHS). We prospectively evaluated whether urinary levels of F₂-IsoP-M and the metabolite (F₂-IsoP-M) to parent compound (F₂-IsoPs) ratio (MPR) were associated with repeatedly measured weight changes over 15-year follow-up in women.

Materials and Methods

Study participants

Study participants were selected from the SWHS, an ongoing large-scale population-based prospective cohort study in seven urban communities of Shanghai, China. Design and methods of the SWHS have been described in detail elsewhere [30]. Briefly, at enrollment between 1997 and 2000, a total of 74,941 women aged 40–70 years were recruited, with an overall response rate of 92.7%. The SWHS was approved by Institutional Review Boards for human research of Vanderbilt University Medical Center and the Shanghai Cancer Institute. Written informed consent was obtained from all study participants.

Participants of the current study were originally selected in two independent nested case-control studies of colorectal and breast cancers within the SWHS [31, 32]. Urinary F₂-IsoPs and F₂-IsoP-M had been measured in these 2 studies, with the same mass spectrometry (MS) based method [33]. Due to the nature of the prospective study design of the two nested case-control studies, all participants were apparently healthy and free of cancer when biological sample collection and exposure assessment were performed. In total, 2066 women were included in this analysis.

Urinary F₂-IsoPs and F₂-IsoP-M measurement

A spot urine sample was collected from each participant at baseline and stored at −80°C until laboratory analyses. Concentrations of F₂-IsoPs and F₂-IsoP-M in urine were determined at Vanderbilt Eicosanoid Core Laboratory using gas chromatography/negative ion chemical ionization mass spectrometry (GC/NICI MS) assays. The analytic method has been described in detail elsewhere [33, 34]. The lower limit of sensitivity was approximately 5 pg. The precision of the assays was ± 6%, and the accuracy was 96%. After standardization by urinary creatinine concentrations, final results were expressed as nanograms per milligram of creatinine (ng/mg Cr).

Weight change and covariate data

Weight measurement was performed by trained interviewers at baseline and repeated thrice during follow-up at year 7, year 10, and year 15. A digital weight scale was used and was calibrated every 6 months. Other anthropometric measurements, including height and circumferences of the waist and hips, were taken at baseline. Each parameter was measured twice with a tolerance error of 1 cm for height, 0.5 cm for circumference measurement, and 1 kg for weight measurement. If the difference between two measurements was greater than the tolerance, a third measurement was then taken. All parameters were estimated as the mean value of the two closest measurements. BMI was calculated as weight in kilograms divided by the square of height in meters. Weight change and percentage of weight change relative to baseline measures were calculated.

Structured questionnaires through in-person interviews were used at baseline to collect detailed information on demographic characteristics, dietary habits, lifestyle factors, and medical history among others [30]. More importantly, energy intake, physical activity and comorbidities, three major contributors to weight change, were assessed at baseline and updated during each of follow-up visits [35–37].

Statistical analysis

Linear mixed models, with a random intercept, were used to estimate associations of urinary F₂-IsoP-M and the MPR with repeated measures of weight change (both weight change in kg and percentage of weight change) over follow-up periods. These associations could be regarded as summary estimates of the overall association between exposure and outcome throughout the entire follow-up period. Moreover, linear regression models were also applied to estimate the association for a given follow-up period. Urinary F₂-IsoP-M concentrations were modeled as categorical variables based on the cutoffs of assay batch-specific quartile distributions, with the lowest quartile serving as the reference group. Tests for trend were conducted by using a Wald test for a variable taking the median concentration of each quartile of the biomarker. Further, urinary F₂-IsoP-M concentrations were also log-transformed and modeled as a continuous variable (per log₂ unit increase), to assess the effect for each doubling of exposure. In addition, we also assessed the MPR, reflecting peroxisomal β-oxidation activity with consideration of the substrate levels. Covariates adjusted for in multivariable models included age, height, education, cigarette smoking, physical activity (time-varying), menopausal status, regular use of vitamin supplements, Charlson comorbidity score (time-varying), and total energy intake (time-varying), as well as use of antibiotics, vitamin supplements and non-steroidal anti-inflammatory drugs (NSAIDs) in the past 24 hours before sample collection. Models for weight change in kg were additionally adjusted for baseline weight.

Data used in this analysis were originally generated in two nested case-control studies of cancers in the SWHS [31, 32]. Though all participants included in this analysis were cancer-free at baseline when biological samples were collected, we examined whether the biomarker and weight change association differed by subsequent risk of cancer. Two analytic approaches were applied by 1) adjusting for cancer type and cancer risk in multivariable models and 2) confining analyses to the control group. All analyses were conducted with R software (version 4.0.4, R Project for Statistical Computing). Two-sided P values of <0.05 were interpreted as statistically significant.

Results

The mean age (standard deviation [SD]) of study participants at baseline was 55.3 (9.2) years. Mean weight was 60.4 (8.9) kg, and mean height was 1.6 (0.1) m. Study participants’ baseline characteristics in relation to urinary F₂-IsoP-M concentrations and the MPR are presented in Table 1. Women with higher F₂-IsoP-M and MPR levels were likely to have higher body weight and BMI. In contrast, F₂-IsoPs levels were not associated with either of them. Compared with the baseline weight, weight measured during follow-up was reduced over time (online suppl. Table S1). As expected, older women tended to experience greater weight loss with increasing age (online suppl. Table S2).

Table 1.

Baseline characteristics of study participants in relation to urinary F₂-isoprostane metabolites

Characteristics	Mean ± SD, or n (%) a	F2-IsoPs b		F2-IsoP-M b		F2-IsoP-M to F2-IsoPs ratio b
		β (SE) c × 100	P value	β (SE) c × 100	P value	β (SE) c × 100	P value
Age (years)	55.3 ± 9.2	−1.33 (0.37)	<0.001	0.61 (0.36)	0.09	1.82 (0.43)	<0.001
Height (m)	1.6 ± 0.1	−111.17 (38.42)	0.004	−170.72 (37.69)	<0.001	−65.42 (43.91)	0.14
Weight (kg)	60.4 ± 8.8	−0.31 (0.22)	0.16	1.06 (0.21)	<0.001	1.44 (0.25)	<0.001
Body mass index (kg/m2)	24.5 ± 3.4	−0.66 (0.54)	0.22	2.65 (0.53)	<0.001	3.48 (0.61)	<0.001
Education, high school and above	781 (37.9)	−4.87 (3.87)	0.21	−7.42 (3.8)	0.05	−2.62 (4.43)	0.56
Cigarette smoking	42 (2.0)	27.20 (12.59)	0.03	19.55 (12.07)	0.11	−6.94 (14.05)	0.62
Postmenopausal	1252 (60.6)	11.33 (6.57)	0.09	8.59 (6.44)	0.18	−1.88 (7.50)	0.80
Charlson comorbidity index	0.2 ± 0.6	4.21 (2.84)	0.14	4.97 (2.78)	0.07	1.58 (3.23)	0.63
Use of antibiotics	109 (5.3)	−7.06 (7.82)	0.37	−2.31 (7.67)	0.76	4.99 (8.93)	0.58
Use of aspirin and other NSAIDs	48 (2.3)	4.29 (12.01)	0.72	−18.13 (11.78)	0.12	−19.36 (13.71)	0.16
Use of vitamin supplements	408 (19.7)	−5.84 (4.62)	0.21	−8.81 (4.52)	0.05	−2.36 (5.27)	0.65
Physical activity (MET, hrs/wk/y)	5.5 ± 11.2	−0.11 (0.16)	0.49	−0.35 (0.16)	0.03	−0.22 (0.18)	0.23
Total energy intake (kcal/d)	1.7 ± 0.4	6.82 (4.46)	0.13	9.14 (4.38)	0.04	2.84 (5.10)	0.58

Note: Total number of participants = 2066. Geometric mean (95% CI) for F₂-IsoPs (ng/mg Cr), 1.59 (1.55, 1.63); for F₂-IsoP-M (ng/mg Cr), 0.58 (0.56, 0.59); for F₂-IsoP-M to F₂-IsoPs ratio (%), 36.04 (35.08, 37.03).

^aMean ± SD were presented for continuous variables; and n (%) were presented for categorical variables.

^bLog-transformed concentrations of biomarkers were treated as dependent variables in the linear regression model, and covariates listed in the table were mutually adjusted for.

^cRegression coefficient (β) and its standard error (SE) estimate were presented.

Table 2 presents baseline urinary levels of F₂-IsoPs, F₂-IsoP-M and MPR in relation to repeatedly measured weight change using mixed-effect models. No association between F₂-IsoPs levels and weight change was observed in this study population. However, a significantly inverse association was found for F₂-IsoP-M. Adjustment for covariates had a minimal effect on the observed estimates, as evidenced in the minimally adjusted model (online suppl. Table S3) vs. the fully adjusted model (Table 2). From the fully adjusted model (Table 2), each doubling in urinary concentrations of F₂-IsoP-M was associated with a reduction in weight by 0.60% (95% CI = −1.10, −0.11). Results were consistent when quartiles of F₂-IsoP-M concentrations were examined. Women in the highest quartile group experienced excess weight loss by 1.33% (95% CI = −2.41, −0.24), compared with those in the lowest quartile group, with P for trend of 0.01. A similar inverse association was found for the MPR (β per doubling = −0.42; 95% CI = −0.84, −0.00; P for trend across quartiles = 0.03). Moreover, an inverse association was also suggested between F₂-IsoP-M and weight change in kg, although it was not statistically significant.

Table 2.

Baseline urinary F₂-isoprostane metabolites in relation to repeatedly measured weight change

Biomarker	Weight change in kg		Percentage of weight change
	β (95% CI) a	P value	β (95% CI) b	P value
F2-IsoPs
Q1	Reference		Reference
Q2	−0.03 (−0.62, 0.55)	0.91	0.09 (−0.92, 1.11)	0.86
Q3	0.51 (−0.08, 1.09)	0.09	1.06 (0.03, 2.08)	0.04
Q4	−0.35 (−0.94, 0.24)	0.25	−0.33 (−1.36, 0.71)	0.54
P for trend c	0.29		0.58
Continuous (log2)	−0.09 (−0.36, 0.17)	0.48	−0.00 (−0.46, 0.45)	0.98
F2-IsoP-M
Q1	Reference		Reference
Q2	0.15 (−0.44, 0.74)	0.62	−0.35 (−1.38, 0.69)	0.51
Q3	−0.24 (−0.85, 0.37)	0.44	−1.26 (−2.32, −0.21)	0.02
Q4	−0.19 (−0.81, 0.44)	0.56	−1.33 (−2.41, −0.24)	0.02
P for trend c	0.38		0.01
Continuous (log2)	−0.14 (−0.43, 0.14)	0.34	−0.60 (−1.10, −0.11)	0.02
F2-IsoP-M to F2-IsoPs ratio
Q1	Reference		Reference
Q2	0.08 (−0.51, 0.68)	0.79	−0.28 (−1.32, 0.76)	0.60
Q3	−0.04 (−0.65, 0.57)	0.90	−1.06 (−2.12, −0.01)	0.05
Q4	−0.06 (−0.68, 0.55)	0.84	−1.09 (−2.16, −0.02)	0.05
P for trend c	0.74		0.03
Continuous (log2)	−0.01 (−0.25, 0.23)	0.93	−0.42 (−0.84, 0.00)	0.05

^aMultivariable adjusted for age, baseline weight, height, education, cigarette smoking, physical activity, menopausal status, regular use of vitamin supplements, Charlson comorbidity score and total energy intake, as well as use of antibiotics, vitamin supplements and NSAIDs in 24 hours before sample collection.

^bMultivariable adjusted for age, height, education, cigarette smoking, physical activity, menopausal status, regular use of vitamin supplements, Charlson comorbidity score and total energy intake, as well as use of antibiotics, vitamin supplements and NSAIDs in 24 hours before sample collection.

^cStatistical tests for trend were conducted by entering the median of each quartile into the model as continuous variables.

We further evaluated whether the association between baseline urinary F₂-IsoP-M and percentage of weight change varied by follow-up periods (Table 3). Increased levels of F₂-IsoP-M and the MPR were both associated with excess risk of weight loss across all three follow-up periods, although only statistically significant for the prolonged period (15 years) of follow-up, suggesting that the relationship became more evident with aging.

Table 3.

Associations between baseline urinary F₂-isoprostane metabolites and percentage of weight change by follow-up periods

Biomarker	At 7-year follow-up		At 10-year follow-up		At 15-year follow-up
	β (95% CI) a	P value	β (95% CI) a	P value	β (95% CI) a	P value
F2-IsoP-M
Q1	Reference		Reference		Reference
Q2	0.15 (−0.99, 1.29)	0.80	0.28 (−1.19, 1.75)	0.71	−0.97 (−2.43, 0.48)	0.19
Q3	−0.84 (−2.00, 0.33)	0.16	0.05 (−1.45, 1.55)	0.95	−1.67 (−3.15, −0.18)	0.03
Q4	−1.18 (−2.37, 0.01)	0.05	−1.35 (−2.89, 0.18)	0.08	−1.61 (−3.17, −0.05)	0.04
P for trend b	0.02		0.04		0.05
Continuous (log2)	−0.50 (−1.05, 0.05)	0.07	−0.44 (−1.14, 0.27)	0.23	−0.82 (−1.53, −0.12)	0.02
F2-IsoP-M to F2-IsoPs ratio
Q1	Reference		Reference		Reference
Q2	0.20 (−0.95, 1.35)	0.73	−0.30 (−1.77, 1.17)	0.69	0.03 (−1.44, 1.50)	0.97
Q3	0.04 (−1.12, 1.19)	0.95	−1.56 (−3.06, −0.06)	0.04	−1.57 (−3.07, −0.07)	0.04
Q4	−0.16 (−1.33, 1.02)	0.79	−0.90 (−2.40, 0.61)	0.24	−1.68 (−3.21, −0.15)	0.03
P for trend b	0.68		0.21		0.01
Continuous (log2)	−0.23 (−0.69, 0.23)	0.33	−0.46 (−1.05, 0.13)	0.12	−0.61 (−1.21, −0.02)	0.04

^aMultivariable adjusted for age at baseline, height, education, cigarette smoking, physical activity, menopausal status, regular use of vitamin supplements, Charlson comorbidity score and total energy intake, as well as use of antibiotics, vitamin supplements and NSAIDs in 24 hours before sample collection.

^bStatistical tests for trend were conducted by entering the median of each quartile into the model as continuous variables.

We thus examined whether the F₂-IsoP-M and weight change association differed by baseline age (online suppl. Table S4). We found that F₂-IsoP-M associated weight loss appeared more pronounced among individuals who were older at baseline, although the P for multiplicative interaction was not statistically significant. A similar age-dependent association was suggested when analyzing for the longest observation at year-15 follow-up (online suppl. Table S5).

We further examined if the association between urinary F₂-IsoP-M levels and weight change was biased by subsequent risk of cancer. We found that it was neither confounded by cancer status nor by cancer types (online suppl. Tables S6 and S7). Moreover, results from analyses confined to the control group were apparently consistent with the results observed in all participants (online suppl. Tables S8 and S9).

Discussion

In this longitudinal analysis, we found that levels of urinary F₂-IsoP-M and the ratio of urinary F₂-IsoP-M to unmetabolized urinary F₂-IsoPs, the MPR, were inversely associated with percentage of weight change, particularly in older women. The inverse association was independent of major determinants to body weight, such as energy intake, physical activity and comorbidities; and it was observed across different follow-up periods, although appearing statistically more significant with prolonged observation.

Weight change can be defined and modelled in two different ways, namely absolute change (weight change in kg) and relative change (percentage of weight change) [38]. The two different measures lend themselves to different research questions and purposes. There is evidence that considering only absolute weight change in population studies may ignore differences in baseline body size [39]. Relative weight change, as a percent change against baseline weight, is commonly used in clinical settings to determine a meaningful change in weight and used in obesity treatment guidelines for weight control [40]. Accordingly, we mainly focused on the relative instead of absolute weight change to take the baseline body size into account, and indeed, only relative weight change (percentage of weight change) was found to be significantly associated with F₂-IsoP-M in this study.

An inverse association between F₂-IsoPs and weight change has been shown in two previous studies [28, 29]. Kanaya, et al. reported, among 366 older women in the Health, Aging, and Body Composition (Health ABC) study, that over 5 years of follow-up, women in the highest vs. lowest tertile of F₂-IsoPs exhibited significant loss of weight [29]. A similar inverse association between F₂-IsoPs concentrations and weight change was reported among 299 participants in the Insulin Resistance Atherosclerosis Study (IRAS) [28]. The present study, to our knowledge, was the largest study of its kind (n = 2066) and, for the first time, used longitudinally repeated measures of weight change during 15-year follow-up in Asian women. We also found a moderate correlation between concentrations of F₂-IsoPs and F₂-IsoP-M (r = 0.3), suggesting that lipid β-oxidation, measured by F₂-IsoP-M, may be one of the mechanisms underlying the association between F₂-IsoPs and weight change observed in previous studies [28, 29].

None of the previous studies have directly evaluated F₂-IsoP-M and whether this metabolite can predict future weight change [28, 29]. In this study, we found that baseline levels of urinary F₂-IsoPs was not associated with subsequent weight change, while levels of its metabolite F₂-IsoP-M were significantly associated with weight loss over time. F₂-IsoP-M is an enzymatic metabolite of F₂-IsoPs formed via a single step of β-oxidation and reduction of the Δ⁵ double bond. It is understood that peroxisomes are responsible for the β-oxidation of prostanoids, including F₂-IsoPs. β-oxidation of very long-chain fatty acids (VLC-FA) and subsequent production of dicarboxylic acid carnitines also occur in the peroxisome [41]. Metabolomic studies identified VLC-FA and dicarboxylic acid carnitines as key variables associated with disease severity in patients with sarcopenia [42]. These findings combined with our results suggest a potential link between sarcopenia and dysregulation of peroxisomal β-oxidation.

Sarcopenia-associated muscle loss and dysregulation of β-oxidation are also linked to mitochondrial dysfunction [43]. Generation of reactive oxygen species and consequential lipid peroxidation and oxidative stress are hallmarks of the disease. It has been shown that plasma F₂-IsoP levels were significantly correlated with circulating biomarkers of skeletal muscle damage [44]. Further, Ma et al. demonstrated that F₂-IsoP-M concentrations were significantly and inversely associated with lean body mass and trunk fat mass, two factors important for the development of aging-related sarcopenia [27]. Peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) is a key transcriptional coactivator that regulates mitochondrial biogenesis, metabolic function, and antioxidant gene expression [45]. Upregulation of PGC-1α has been shown to decrease levels of mitochondrial F₂-IsoPs in a mouse model of prolong immobilization, which is a risk factor for sarcopenia in older adults [46]. Further investigations into mitochondrial dysfunction, lipid oxidation and metabolism in relation to unintentional aging-related weight loss are needed.

This study has a few methodological strengths, including its prospective nature, a large sample size, and use of the best approach currently available for biomarker measurement. Compared to plasma, urine is an ideal biological material for the measurement of F₂-IsoPs and their metabolites because it is low in lipid contents and hence is less likely to be influenced by lipid autoxidation during sampling and storage. Moreover, the MS-based method is considered as the gold standard for the quantification of F₂-IsoPs and F₂-IsoP-M due to its demonstrated high sensitivity and specificity [19, 32]. Importantly, F₂-IsoP-M is not subject to local renal production [33], thus considering as a good systemic in-vivo biomarker. Another distinguished feature of our study is the availability of longitudinally repeated measurements of weight change, which allowed us to explore trajectory of weight change in relation to lipid metabolism. Further, all anthropometric measurements were performed by trained interviewers, which avoided misclassification arising from the self-reported data.

There are some limitations/concerns in our study. One potential concern was whether a single measure in urine can represent long-term exposure of F₂-IsoPs and F₂-IsoP-M reasonably well. We have carefully evaluated their intra-person variations over time and found that urinary levels of F₂-IsoPs and F₂-IsoP-M remain rather stable, with intraclass correlation coefficient (ICC) being 0.69 for F₂-IsoPs and 0.76 for F₂-IsoP-M over time [47]. These ICCs are very similar to those found for serum cholesterols, which is commonly acknowledged that a single measure represents an individual’s long-term exposure reasonably well and is well accepted as a risk predictor for cardiovascular disease [48].

Another concern is that we could not totally rule out the possibility of residual confounding owing to unmeasured or inaccurately measured covariates. However, data on a large range of covariates were available, including demographic and lifestyle factors and lipid peroxidation- and inflammation-related exposures/conditions. More importantly, this study measured both the outcome variable (weight change) and three major determinants of weight change (energy intake, physical activity and comorbidities) at baseline and repeatedly thrice during follow-up, which enabled us to comprehensively control for these major confounding factors. On the other hand, because all participants in this study were middle-aged and older Chinese women, although a study conducted in a single racial population may reinforce the internal validity of the findings, their generalizability to other populations may be of concern.

This study used the data originally generated in two independent nested case-control studies of cancers in the SWHS. Although, because of the prospective nature of the two studies, all participants were free of cancer at baseline when urine samples were collected for biomarker measurement, there were concerns if the association between F₂-IsoP-M and weight change might be biased by subsequent risk of cancer. However, we found that neither additionally adjusting for cancer risk nor confining analyses to the control group materially changed the results.

In summary, this study found the first piece of evidence that increased levels of F₂-isoprostane metabolite were associated with excess risk of weight loss over 15-year follow-up, suggesting that this biomarker could be a novel indicator for weight loss among elderly women. Further investigations are warranted to determine the role of oxidative damage and metabolic processes, including peroxisomal β-oxidation, in aging-related weight change.

Abbreviation List:

BMI

body mass index

CI

confidence interval

F₂-IsoPs

F₂-isoprostanes

F₂-IsoP-M

2,3-dinor-5,6-dihydro-15-F_2t-IsoP

GC/NICI-MS

gas chromatography/negative ion chemical ionization mass spectrometry

ICC

intraclass correlation coefficient

MPR

metabolite to parent ratio

MS

mass spectrometry

NSAID

non-steroidal anti-inflammatory drug

SD

standard deviation

SWHS

Shanghai Women’s Health Study

Data Availability Statement

All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author.