Physical activity and cancer survival: the role of peripheral blood immune and inflammatory markers in reducing mortality—a retrospective cohort study based on NHANES 1999–2018

Introduction

Cancer remains one of the most prevalent noncommunicable diseases worldwide and a leading cause of mortality, accounting for approximately 10 million deaths in 2022 (one in six deaths) and posing an increasingly heavy healthcare burden globally (1). Over the past two decades, advances in medical technology have improved treatment outcomes and post-treatment survival for many cancer patients. Consequently, greater attention has been directed toward identifying modifiable lifestyle factors that may further prolong and enhance survival.

Major clinical guidelines have recognized physical activity as an important component of cancer survivorship care. The American Cancer Society (ACS) guideline on nutrition and physical activity for cancer survivors recommends regular physical activity to improve quality of life and reduce the risk of recurrence and mortality (2). Similarly, the American Society of Clinical Oncology (ASCO) clinical practice guideline recommends aerobic combined with resistance exercise as an adjunctive intervention during active cancer treatment, as it significantly reduces fatigue, improves cardiopulmonary function, muscle strength, and quality of life, with particularly robust evidence in breast, prostate, lung, and colorectal cancers (3). However, although physical activity has been shown to benefit survival in cancer patients, the ACS guideline points out that most research has focused on all-cause mortality, and the effects of physical activity on cancer-specific and cardiovascular mortality remain unclear. Furthermore, the potential underlying biological mechanisms are also poorly understood (2). This evidence gap underscores the need for large-scale, real-world studies examining the association between physical activity and mortality in diverse cancer survivor populations.

Research has shown that physical activity is associated with reduced oxidative stress, improved immune homeostasis, and lower levels of persistent inflammation (4). Some studies have also suggested that increasing exercise duration and moderately raising exercise intensity may be linked to enhanced lymphocyte antitumor activity, reduced immunosuppressive effects, and better responses to anticancer treatments, which could be associated with improved prognosis (5-10).

Systemic inflammation is generally assessed via peripheral blood immune and inflammatory markers (PBIMs). Commonly used indices include the systemic immune-inflammation index (SII), neutrophil-lymphocyte ratio (NLR), monocyte-lymphocyte ratio (MLR), neutrophil percentage-albumin ratio (NPAR), C-reactive protein-albumin-lymphocyte (CALLY) index, advanced lung cancer inflammation index (ALI), and prognostic nutritional index (PNI). These indices, derived from routine blood tests, offer a non-invasive and cost-effective way to evaluate systemic immune and inflammatory status and have been associated with cancer risk and prognosis in previous studies (11-13).

Given that physical activity may influence inflammatory profiles and that inflammatory markers have been linked to cancer survival, it is plausible that PBIMs play a mediating role in the association between physical activity and mortality in cancer survivors. However, whether and to what extent these blood markers explain the physical activity-survival relationship remains largely unexplored, particularly in large, real-world cancer survivor populations. Therefore, in this study, we aimed to explore this potential indirect pathway while acknowledging that our analysis is exploratory in nature.

The Centers for Disease Control and Prevention (CDC) conducted the National Health and Nutrition Examination Survey (NHANES), a study based on population data that includes real-world information from the long-term monitoring of cancer survivors. It contains comprehensive data on lifestyle, exercise habits, full blood counts, and inflammatory markers, with longitudinal mortality follow-up through linkage to the National Death Index.

We conceptualize PBIMs as potential mediators based on biological plausibility and supporting literature. First, regular physical activity has been shown to reduce systemic inflammation, such as C-reactive protein and interleukin-6, and improve lymphocyte profiles, while these inflammatory markers have been independently associated with cancer prognosis (4,7-13). Second, in NHANES, physical activity was assessed over the 30 days preceding blood collection, whereas PBIMs were measured from blood drawn on the examination day, supporting the directional assumption that past activity may influence current blood parameters. Third, prior studies have provided empirical evidence supporting the mediating role of inflammatory and metabolic pathways in the association between physical activity and survival outcomes. For example, a recent NHANES-based study reported that leisure-time and transportation-related moderate-to-vigorous physical activity were inversely associated with systemic inflammatory markers (14). Although our cross-sectional design precludes causal inference, these lines of evidence collectively justify the exploration of PBIMs as potential mediators in the physical activity–mortality association.

This study, therefore, aims to provide real-world evidence on the association between physical activity and survival in cancer survivors, and to explore the potential role of PBIMs in this association, with the ultimate goal of informing future research on strategies to improve long-term survival. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0462/rc).

Methods

The NHANES is a national survey that provides a cross-sectional view of the noninstitutionalized civilian population in the United States (US) (15) and is composed of data collected in 2-year cycles since 1999. NHANES employs a stratified, multistage probability sampling design to select a nationally representative sample. The sampling procedure consists of four stages: (I) selection of primary sampling units (counties or groups of contiguous counties); (II) selection of segments (groups of dwelling units) within primary sampling units; (III) selection of dwelling units within segments; and (IV) selection of sample persons within dwelling units. Oversampling was implemented to increase the reliability of estimates for specific subgroups, including racial/ethnic minorities, low-income populations, and older adults. Detailed descriptions of the sample design are available in the NHANES analytic guidelines published by the National Center for Health Statistics. Data from the NHANES are derived from comprehensive interviews that elicit information on demographics, socioeconomic conditions, eating patterns, and various health factors. Furthermore, thorough examinations, including medical tests and blood marker assessments, are performed by skilled healthcare providers to generate clinical data. All participants in the NHANES study provided written informed consent, and the protocol was approved by the National Center for Health Statistics (NCHS) Ethics Review Board. This modeling study did not require a review, as it used published datasets that were de-identified and contained no personal information. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Study population

Data for this study were collected from the NHANES database, which consists of 10 cycles of data collection conducted between 1999 and 2018. Information on cancer diagnoses was ascertained during the NHANES data collection process using questionnaire item MCQ-220. Participants were asked whether they had ever been diagnosed with cancer or any malignancy by a doctor or health professional. A total of 113,249 participants aged ≥20 years were initially considered from 10 NHANES cycles [1999–2018]. After sequentially excluding participants with missing cancer diagnosis data (n=108,083), those with missing physical activity data (n=906), those with incomplete blood test indicators (n=1,491), and those missing mortality outcome data (n=1), a final analytical cohort of 2,768 cancer survivors was obtained.

Assessment of physical activity

Each participant completed a physical activity questionnaire (PAQ). The questionnaire included inquiries pertaining to all physical activities undertaken within the preceding 30 days. The recorded data included the type of activity, how long it lasted, its intensity, and how often it had been performed in the previous 30 days. A metabolic equivalent (MET) is the comparison of the work metabolic rate to a standard resting metabolic rate corresponding to quiet sitting (1.0 MET =4.184 kJ·kg⁻¹·h⁻¹) and can be used to classify the intensity of physical activities and facilitate comparisons across studies. The energy expenditure (e.g., in MET-minutes or kcal) can be estimated by combining an activity’s MET value with its duration and an individual’s body weight (16). To calculate MET minutes per 30 days (MET min/30 days) for each activity, MET scores were multiplied by the average session duration and the frequency of performance over the previous 30 days. The total weekly MET minutes were calculated by summing the MET minutes across all activities over 30 days and dividing them by 4.29. According to the US national physical activity guidelines (17), participants were categorized into a low physical activity group and a high physical activity group (low physical activity: <500 MET/week; high physical activity: ≥500 MET/week). Approximately 500 MET is equivalent to 3.3 hours of slow walking or 71 minutes of freestyle swimming per week.

The definition and calculation of PBIMs

In this study, inflammatory indices were determined based on specific indicators present in the peripheral blood. All PBIMs were derived from laboratory data collected during the NHANES Mobile Examination Center examination. Complete blood counts, including neutrophil, lymphocyte, monocyte, and platelet counts, were measured using a Beckman Coulter DxH 800 automated hematology analyzer in the Mobile Examination Center. The complete blood counts methodology is based on the Beckman Coulter principle of counting and sizing, combined with volume, conductivity, and scatter technology for 5-part differential. High-sensitivity C-reactive protein was measured using a two-reagent immunoturbidimetric system on a Beckman UniCel DxC 600 Synchron or DxC 660i Synchron clinical chemistry analyzer, with latex particles coated with mouse anti-human CRP antibodies. Serum albumin was measured using the bromcresol purple method. Detailed laboratory procedures, including quality assurance and quality control protocols, are publicly available from the NHANES Laboratory Methods files. Included among these indicators were total white blood cell count (×10⁹/L), neutrophil count (×10⁹/L), lymphocyte count (×10⁹/L), monocyte count (×10⁹/L), platelet count (×10⁹/L), albumin level (g/L), C-reactive protein level (mg/L), lymphocyte percentage, monocyte percentage, and neutrophil percentage—all of which were obtained from the laboratory data section of the database. The following are the formulas for the markers:

SII=[plateletcount(×109/L)×neutrophilcount(×109/L)]lymphocytecount(×109/L)[1]

NLR=neutrophilcount(×109/L)lymphocytecount(×109/L)[2]

MLR=monocytecount(×109/L)lymphocytecount(×109/L)[3]

NPAR=neutrophilpercentage(%)albuminlevel(g/dL)[4]

ALI=lymphocytepercentage(%)×totalWBC(×109/L)100[5]

CALLY=albuminlevel(g/dL)×lymphocytecount(×109/L)C-reactiveproteinlevel(mg/L)×10[6]

PNI=10×albuminlevel(g/dL)+0.005×lymphocytecount(×109/L)[7]

Characteristics related to sociodemographics, lifestyle practices, and enduring medical conditions

Self-reported sociodemographic characteristics included the following: sex (male or female), age (<40, 40 to <65, and ≥65 years), race (Mexican American, non-Hispanic Black, non-Hispanic White, and other races), educational attainment (< high school, high school, and > high school), marital status (married, single, and never married), and household poverty-income ratio (PIR; <1.3, 1.3 to <3.5, and ≥3.5). The lifestyle factors considered were smoking history (nonsmoking and smoking) and history of alcohol consumption (nondrinking and drinking). Hypertension was reported in all participants. The diagnosis was received from a healthcare professional or determined by NHANES-measured blood pressure [≥140 mmHg (systolic) or ≥90 mmHg (diastolic)]. A person was diagnosed with hyperlipidemia if they met any of the following criteria: total cholesterol ≥200 mg/dL, triglyceride ≥150 mg/dL, and low-density lipoprotein cholesterol ≥130 mg/dL. High-density lipoprotein cholesterol levels were considered low if they were ≤40 mg/dL in males and ≤50 mg/dL in females. Individuals taking lipid-lowering drugs were also classified as hyperlipidemic. Diabetes mellitus was diagnosed in an individual if they satisfied any one of the following criteria: a documented history of diabetes mellitus, insulin therapy, use of diabetes medications for blood glucose reduction, hemoglobin A1c levels ≥6.5%, fasting glucose levels ≥126 mg/dL, and 2-hour postprandial glucose levels ≥200 mg/dL.

Analysis of mortality

The death status of the NHANES participants was obtained from the National Death Index Linked Mortality Files. In our study, the primary outcomes of interest were all-cause, cancer, and cardiovascular mortality. All-cause mortality included deaths from any cause, cardiovascular mortality included deaths due to heart and cerebrovascular diseases, and cancer mortality was classified according to the International Classification of Diseases, 10^th Revision (ICD-10) code C00-C97. Cardiovascular mortality was defined using ICD-10 codes I00–I09, I11, I13, I20–I51. The follow-up period spanned from the date of initial diagnosis to either the date of death or the end of the study on December 31, 2019, depending on which came first.

Statistical analysis

As NHANES employed a complex, multistage probability sampling design to select representative participants, data analysis included primary sampling units, strata, and sample weights. Via the survey package in R version 4.5.2 (The R Foundation for Statistical Computing, Vienna, Austria), weighted analyses were applied to develop estimates representative of the entire nation. This method guaranteed that the results could be applied to the noninstitutionalized US population and prevented the exaggeration of statistical significance. According to the NHANES guidelines, the choice of weighting variables should prioritize the accurate representation of small-population subgroups through the application of the correct weights. The study population was classified into a control group and an experimental group according to the level of physical activity. As all continuous variables had skewed distributions, they are represented as the median with interquartile range (IQR). Meanwhile, categorical variables are expressed as counts and percentages. The Mann-Whitney test was used to evaluate differences between groups for continuous variables, while the Rao-Scott Chi-squared test was applied for categorical variables owing to the complex survey design of the NHANES. To meet the assumptions of parametric tests in the analysis of the relationship between physical activity and PBIMs, all PBIMs exhibiting a skewed distribution were log₁₀-transformed and analyzed via analysis of covariance (ANCOVA).

We used multivariable Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) to determine the association of physical activity and PBIMs with all-cause and cause-specific mortality. Three sequential models were fitted: a crude model (Model 1), a model adjusted for demographic factors (Model 2: age, sex, race, education, PIR, and marital status), and a fully adjusted model (Model 3: Model 2 plus behavioral and clinical covariates, including smoking, alcohol use, diabetes, hypertension, and hyperlipidemia). Given their skewed distributions, all PBIMs were log₁₀-transformed prior to analysis, and HRs for these markers corresponded to a one-order-of-magnitude (10-fold) change in their original concentration.

The relationship between log-transformed PBIMs and mortality was analyzed via restricted cubic splines in fully adjusted Cox models. The likelihood ratio test was employed to evaluate nonlinearity. For relationships that were significantly nonlinear, we characterized the association by identifying inflection points and performing threshold analysis using piecewise regression. To assess the reliability of the primary results, we conducted sensitivity analyses by excluding mortality events within the first 2 years of follow-up and performed stratified analyses across subgroups defined by age, sex, race, and comorbidity history.

Mediation analysis was performed to estimate the proportion of the association between physical activity and mortality mediated by the PBIMs. The follow-up time was determined from the baseline survey to the first occurrence of death, loss to follow-up, or December 31, 2019. All analyses incorporated sampling weights to reflect the complex sampling design of NHANES.

All statistical analyses were performed with R software version 4.5.2. The survey R package was employed for all analyses, incorporating NHANES sampling weights, and the mediation R package was used for causal mediation analysis. A two-sided P value of less than 0.05 was considered to indicate statistical significance.

Results

Population characteristics

The study comprised 2,768 individuals, 56.86% of whom were female, after those who did not meet the inclusion criteria were excluded. The weighted mean age of the participants was 61.93 [standard error (SE), 0.38; range, 20–85] years, and the median follow-up time was 100.00 (range, 1–249) months. The participant characteristics are presented in Table 1. Among them, there were 1,573 (56.83%) with high physical activity. Participants exhibiting low levels of physical activity were more likely to be older (54.67%), married (62.23%), educated (59.37), non-Hispanic White (84.43%), female (61.95%), and middle-income status (41.30%) (all P values <0.05). Smoking history was not significantly different between the groups (P>0.05).

High physical activity, as compared to low physical activity, was correlated with a higher PNI (51.50 vs. 51.00, P=0.046) and CALLY (4.81 vs. 3.25, P<0.001); moreover, high physical activity was correlated with a lower SII (486.44 vs. 541.05, P<0.001), NPAR (14.04 vs. 14.45, P<0.001), NLR (2.09 vs. 2.23, P=0.003), and MLR (0.29 vs. 0.31, P=0.002). The results are presented in Table 2.

Association between physical activity and mortality

Over a mean follow-up of 8.33 years, 831 deaths were recorded, including 255 cancer-related deaths and 576 noncancer deaths. We used the low physical activity group as the reference group and employed three progressively adjusted models. Prior to any adjustment, physical activity was associated with all-cause mortality in cancer survivors (HR =0.59; 95% CI: 0.50–0.70; P<0.001), cardiovascular mortality (HR =0.61; 95% CI: 0.43–0.86; P=0.005), and cancer-specific mortality (HR =0.75; 95% CI: 0.57–0.98; P=0.04), while a higher level of physical activity was associated with a reduced risk of mortality. Based on these results, we plotted the Kaplan-Meier survival curve for Model 1 (Figure 1). However, after adjustments were made for age, sex, race, educational, PIR, and marital status (Model 2), physical activity in cancer survivors was only associated with mortality from all causes (HR =0.63; 95% CI: 0.55–0.73; P<0.001) and cardiovascular mortality (HR =0.66; 95% CI: 0.47–0.92; P=0.02). Even with additional adjustments for smoking, alcohol use, diabetes, hypertension, and hyperlipidemia (Model 3), the results remained robust (Table 3). Despite multivariable adjustment for baseline differences, residual confounding cannot be excluded and may affect the observed associations.

Figure 1 Kaplan-Meier survival analysis by physical activity level in patients with cancer. Kaplan-Meier curves comparing survival probabilities between patients with low and high physical activity levels for (A) all-cause, (B) cardiovascular, and (C) cancer-specific mortality. Log-rank P values indicate statistically significant differences between groups for all endpoints. Numbers at risk at specified time points are shown below each graph.

Subgroup analyses for the association between physical activity and mortality

Based on the established association between physical activity and reduced mortality risk, we conducted comprehensive subgroup analyses to clarify the potential effect modifications. Organized by age, sex, race, smoking habits, alcohol consumption, and comorbid conditions such as high blood pressure, diabetes, and high lipid levels, these analyses revealed consistent protective effects of higher physical activity levels against all-cause and cardiovascular mortality across all subgroups. Formal tests for interaction yielded no statistically significant results (all P values >0.05), indicating that the beneficial association between physical activity and survival was homogeneous across these key demographic and clinical characteristics. The forest plot in Figure 2 provides a comprehensive depiction of the results from the subgroup analyses.

Figure 2 Forest plot of the subgroup analysis for the association between PA and all-cause mortality. The forest plot presents the results of a subgroup analysis examining the association between high levels of PA (high PA) and risk of all-cause mortality compared to low levels of PA (low PA) among 2,768 patients with cancer. P: P value for the HR within the specific subgroup; P for interaction: P value for the interaction test, indicating whether the effect of PA differs significantly across the categories of a given subgroup (e.g., male vs. female under “sex”). CI, confidence interval; HR, hazard ratio; PA, physical activity.

Correlation between physical activity and log-transformed PBIMs

Our data suggest that exercise is associated with improved survival in cancer survivors. However, whether exercise can improve the immune and inflammatory status of patients, thereby affecting their survival, remains unclear, and thus, we examined the role of common PBIMs in this process. All peripheral blood inflammatory indicators were log-transformed. In the unadjusted model (Model 1), with low physical activity serving as the reference, high physical activity was negatively correlated with the inflammatory indicators logSII (β=−0.10; 95% CI: −0.15 to −0.05; P<0.001), logMLR (β=−0.07; 95% CI: −0.11 to −0.03; P<0.001), logNLR (β=− 0.09; 95% CI: −0.14 to −0.04; P=0.001), and logNPAR (β=−0.05; 95% CI: −0.07 to −0.03; P<0.001) but positively correlated with logCALLY (β=0.39; 95% CI: 0.27 to 0.52; P<0.001) and logPNI (β=0.02; 95% CI: 0.01 to 0.03; P=0.03), whereas no correlation was observed with logALI (Table 4). Except for logPNI and logALI, all inflammatory indicators remained consistent in Model 2 (adjusted for age, sex, race, education, PIR, and marital status) and Model 3 (adjusted for history of smoking, alcohol use, diabetes, hypertension, and hyperlipidemia). No relationship was observed between the logPNI and physical activity after adjustments were made for age and sex.

Associations of PBIMs with mortality

Having identified several inflammatory markers responsive to physical activity levels, we investigated their individual prognostic value for all-cause and cause-specific mortality. First, we determined the association between logPBIM levels and all-cause mortality using a restricted cubic spline. In the adjusted Model 3, the immune markers exhibited a nonlinear relationship with mortality risk among cancer survivors (Figure 3).

Figure 3 Restricted cubic spline analysis of peripheral blood immune/inflammatory markers and all-cause mortality. (A) LogPNI. (B) LogSII. (C) LogNPAR. (D) LogNLR. (E) LogALI. (F) LogMLR. (G) LogCALLY. The association was evaluated using multivariable-adjusted restricted cubic spline analysis (four knots at the 5^th, 35^th, 65^th, and 95^th percentiles). The solid red line represents the estimated HRs, and the shaded area denotes the 95% CI. The model was adjusted for age, gender, race, education, marital status, PIR, smoking status, alcohol use, diabetes, hypertension, and hyperlipidemia. ALI, advanced lung cancer inflammation index; CALLY, C-reactive protein-albumin-lymphocyte; CI, confidence interval; HR, hazard ratio; MLR, monocyte-lymphocyte ratio; NLR, neutrophil-lymphocyte ratio; NPAR, neutrophil percentage-albumin ratio; PIR, poverty-income ratio; PNI, prognostic nutritional index; SII, systemic immune-inflammation index.

Based on the confirmed nonlinear dynamics, we characterized the threshold effects of the three salient markers using piecewise regression analysis. LogNPAR demonstrated an exceptionally strong J-shaped association, with levels above its inflection point (2.561) associated with a dramatic 953% increase in mortality risk per unit (HR =10.53; 95% CI: 5.33–20.81). LogMLR showed the second strongest effect, with a 126% elevated risk beyond its threshold. In contrast, the emerging marker, logCALLY, exhibited a protective association, with an inflection point of 0.055. Values below this threshold were associated with a significant 34% risk reduction (HR =0.66; 95% CI: 0.54–0.82), whereas above it, a more modest 13% risk reduction was observed (HR =0.87; 95% CI: 0.79–0.95). These quantified risk relationships are visually represented in the threshold effect diagrams in Figure 4. The complete threshold analysis results for all the inflammatory markers are presented in Table 5.

Figure 4 Threshold curves illustrating the associations between seven immune/inflammatory markers indices and all-cause mortality. (A) LogMLR. (B) LogNLR. (C) LogCALLY. (D) LogPNI. (E) LogALI. (F) LogNPAR. (G) LogSII. ALI, advanced lung cancer inflammation index; CALLY, C-reactive protein-albumin-lymphocyte; HR, hazard ratio; MLR, monocyte-lymphocyte ratio; NLR, neutrophil-lymphocyte ratio; NPAR, neutrophil percentage-albumin ratio; PNI, prognostic nutritional index; SII, systemic immune-inflammation index.

PBIMs as mediators for the association of physical activity with mortality

We further investigated how PBIMs influence the relationship between physical activity and survival outcomes in patients with cancer. After adjustments were made for all covariates, logCALLY and logMLR were found to exert a relatively strong mediating effect, with the mediation proportion of logCALLY and logMLR being 8.95% and 10.49%, respectively. The logNPAR and logNLR also exhibited a significant mediation effect, accounting for 7.29% and 8.53% of the total effect, respectively. In contrast, no significant mediation effects were observed for logSII or logPNI.

Discussion

This study analyzed a large, nationally representative cohort of US cancer survivors and found a significant inverse association between higher physical activity levels and the risk of all-cause and cardiovascular mortality. More importantly, the findings suggest a potential indirect pathway, as PBMIs, most notably the CALLY index and MLR, were found to partially mediate this protective association. To the best of our knowledge, this is the first study to investigate the influence of PBIMs on the link between physical activity and survival outcomes in a large, nationally representative group of cancer survivors. Our results are consistent with the hypothesis that physical activity may confer survival benefits through pathways involving PBIMs.

Our principal finding that increased physical activity is correlated with lower all-cause mortality is strongly supported by extensive epidemiological and clinical studies. For instance, the landmark CHALLENGE trial established structured exercise as a cornerstone therapy capable of improving both disease-free and overall survival in patients with colon cancer (18). Furthermore, meta-analyses of randomized controlled trials have confirmed that exercise significantly reduces mortality risk among cancer survivors (3,7,19,20). The consistency of this association across all the demographic and clinical subgroups in our study, with no significant interaction effects, strengthens the validity and generalizability of these findings. Notably, although we observed a significant reduction in all-cause mortality, the associations with both cardiovascular disease-specific mortality and cancer-specific mortality were not significant after full adjustment. Our findings were partially inconsistent with a previous study (21) and may be linked to the heterogeneous nature of the study population and the specific covariates selected in our analysis. Moving beyond a confirmation of this established relationship, our research provides preliminary evidence for the involvement of PBIMs and highlights the influence of exercise on systemic inflammation.

Although high physical activity was associated with reduced all‑cause and cardiovascular mortality in the fully adjusted model, the association with cancer‑specific mortality did not reach statistical significance (HR =0.89; 95% CI: 0.64–1.23), and the cardiovascular mortality association, while still significant, was attenuated compared with the unadjusted model. Several factors may explain these findings. First, clinical heterogeneity (diverse cancer types, stages, and treatments) in our survivor cohort could have diluted cause‑specific effects. Second, competing risks may play a role, as physical activity might exert stronger effects on cardiovascular health than on cancer progression in long‑term survivors. Third, residual confounding from unmeasured variables (e.g., cancer stage, performance status, comorbidities) cannot be excluded. The CI for cancer‑specific mortality is compatible with a modest protective effect, and the lack of statistical significance should not be overinterpreted as evidence of no effect. Future studies with larger cause-specific event numbers and more detailed clinical characterization are needed to clarify these associations.

The most salient finding of our study was that CALLY and MLR partially mediated the association between physical activity and mortality, with NLR and NPAR also contributing to this mediation effect. These mediators can be conceptually divided into those that incorporate nutritional status and those that primarily reflect the immune cell balance. The CALLY index, which combines nutritional status (albumin), immune competence (lymphocytes), and systemic inflammation (C-reactive protein), may serve as a relevant correlate in the pathway from physical activity to survival. Involuntary weight loss is a significant concern for patients with cancer and often serves as a predictor of poor prognosis (22). Our study identified that albumin-related metrics were associated with the relationship between physical activity and survival in patients with cancer, although direct evidence for exercise-enhancing albumin synthesis remains lacking. A multicenter cohort study based on the Investigation on Nutritional Status and its Clinical Outcome of Common Cancers project demonstrated that CALLY functions as a standalone positive predictor for the overall survival of patients with esophageal cancer, exhibiting superior predictive accuracy and clinical value as compared to the NLR, platelet-lymphocyte ratio (PLR), SII, and PNI (23). Our findings are supported by accumulating evidence that CALLY serves as a robust prognostic indicator across various cancer types, with higher values consistently predicting improved survival outcomes (24-28). This finding raises the possibility that part of the survival advantage of physical activity could be related to improvements in inflammation and nutritional status. This aligns with the known physiological concept that exercise contributes to the reduction of systemic C-reactive protein levels (28) and modulates lymphocyte populations. Specifically, exercise stimulates and activates cytotoxic T lymphocytes (CTLs) (8,29,30) and natural killer (NK) cells (31), promoting their movement into the tumor microenvironment. Similarly, NPAR, an albumin marker, demonstrated a significant mediating effect in our study. The regular participation of albumin in both CALLY and NPAR highlights the possible impact of exercise in alleviating cancer-related malnutrition and cachexia, which are closely associated with systemic inflammation and unfavorable outcomes.

A reduced lymphocyte-monocyte ratio (LMR) is often related to a less optimistic prognosis and heightened risk of tumor recurrence (32,33). We investigated the inverse MLR and characterized its nonlinear relationship with the survival of patients with cancer. Our findings indicate that a higher MLR is significantly associated with an increased risk of death from any cause, consistent with the established prognostic value of LMR. Furthermore, we identified a significant inverse association between physical activity and MLR, suggesting that the association between exercise and survival might be partially explained by reductions in the inflammatory ratio. Elevated NLR has also been shown to be associated with poor prognosis in patients with cancer (34,35). Our database study corroborated this result. Furthermore, we found a significant negative correlation of physical activity with MLR and NLR, an observation similar to that reported in a previous study on exercise and immune inflammation based on the NHANES database (14). This suggests that physical activity can improve survival outcomes by lowering inflammatory ratios.

The results of this study may have implications for future research. CALLY, NPAR, NLR, and MLR, which are readily available and cost-effective blood-based indices, could be explored as potential tools in long-term follow-up protocols for cancer survivors. Because these indices are derived from routine blood tests, they require no additional healthcare spending, rendering them a fiscally sustainable option for public health and insurance systems. These indices might be used to evaluate inflammatory and nutritional status, potentially generating hypotheses for future studies on personalized physical activity interventions. However, any clinical application would require validation in prospective longitudinal studies. Clinicians may consider these indicators as auxiliary information for prognostic evaluation, but their routine use is not yet supported by the current evidence. Their simplicity and accessibility make them particularly advantageous in resource-limited healthcare settings for research purposes.

This study has several significant advantages. First, it used the NHANES database, which is a nationally representative cohort with a complex sampling framework, and our results are particularly applicable to the noninstitutionalized US population of cancer survivors. Second, the application of formal mediation analysis allowed us to move beyond establishing an association, and we further quantified the specific proportion of the survival benefit as explained by objective, routinely collected inflammatory biomarkers, thus providing exploratory evidence that may inform future studies on the pathways involved. Finally, the concurrent evaluation of a panel of novel and established PBIMs, coupled with an examination of their nonlinear relationship with mortality, offers a more detailed characterization of the associations between physical activity and survival.

Several limitations should be considered. The single-time-point assessment of physical activity using a 30-day recall may not capture long-term patterns; any non-differential misclassification would likely bias HRs toward the null, making our findings conservative. In addition, self-reported physical activity is subject to recall and social desirability biases, and the dichotomization at 500 MET‑min/week oversimplifies the dose-response relationship, precluding differentiation between activity intensity or type. Furthermore, the broad definition of cancer survivorship (self-reported any cancer) without detailed clinical data (stage, histology, treatment, time since diagnosis) introduces clinical heterogeneity and potential residual confounding; analyses by specific cancer type were not feasible due to incomplete coding and small sample sizes. Moreover, because physical activity and PBIMs were measured concurrently (with activity recalled for the prior 30 days), the temporal ordering required for causal mediation cannot be firmly established, and reverse causality cannot be excluded. Consequently, all mediation findings are exploratory and hypothesis-generating, not evidence of causal pathways. Finally, as an observational study, causality cannot be inferred despite extensive covariate adjustment. Future prospective studies with repeated exposure measurements, objective activity monitors, and detailed clinical characterization are needed to confirm and extend our findings.

Conclusions

Based on an analysis of a large, nationally representative US cohort, this study found that higher physical activity levels were associated with reduced all-cause and cardiovascular mortality among cancer survivors. In addition, our exploratory mediation analysis suggested that PBIMs, particularly the CALLY index and MLR, may partially explain this association. These findings provide preliminary evidence that readily available inflammatory markers could be further investigated as potential tools in survivorship research. However, owing to the observational design and cross-sectional assessment of physical activity and PBIMs, causality cannot be inferred, and the results should be considered hypothesis-generating.

Acknowledgments

We gratefully acknowledge the participants and investigators of the NHANES database for providing the data used in this study. We also thank all individuals who contributed to the completion of this research.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0462/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0462/prf

Funding: This work was supported by funding from the Shandong Provincial Natural Science Foundation (No. ZR2022MH103), the Special Fund for Clinical Research of the Wu Jieping Medical Foundation (No. 320.6750.2023-19-5), and the Special Fund for Clinical Research of the Shandong Provincial Medical Association (No. YXH2022ZX02202).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0462/coif). All authors report funding support from the Shandong Provincial Natural Science Foundation (No. ZR2022MH103), the Special Fund for Clinical Research of the Wu Jieping Medical Foundation (No. 320.6750.2023-19-5), and the Special Fund for Clinical Research of the Shandong Provincial Medical Association (No. YXH2022ZX02202). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All participants in the NHANES study provided written informed consent, and the protocol was approved by the National Center for Health Statistics (NCHS) Ethics Review Board. This modeling study did not require a review, as it used published datasets that were de-identified and contained no personal information. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Rock CL, Thomson CA, Sullivan KR, et al. American Cancer Society nutrition and physical activity guideline for cancer survivors. CA Cancer J Clin 2022;72:230-62. [Crossref] [PubMed]
3. Ligibel JA, Bohlke K, May AM, et al. Exercise, Diet, and Weight Management During Cancer Treatment: ASCO Guideline. J Clin Oncol 2022;40:2491-507. [Crossref] [PubMed]
4. Gassner M, Balaskovits P, Melchior P, et al. Reactive oxygen species and IL-6 responses to resistance-based HIIT: A time-dependent analysis. Free Radic Biol Med 2026;242:250-7. [Crossref] [PubMed]
5. Ahmadi MN, Clare PJ, Katzmarzyk PT, et al. Vigorous physical activity, incident heart disease, and cancer: how little is enough? Eur Heart J 2022;43:4801-14. [Crossref] [PubMed]
6. Morishita S, Hamaue Y, Fukushima T, et al. Effect of Exercise on Mortality and Recurrence in Patients With Cancer: A Systematic Review and Meta-Analysis. Integr Cancer Ther 2020;19:1534735420917462. [Crossref] [PubMed]
7. Friedenreich CM, Stone CR, Cheung WY, et al. Physical Activity and Mortality in Cancer Survivors: A Systematic Review and Meta-Analysis. JNCI Cancer Spectr 2020;4:pkz080. [Crossref] [PubMed]
8. Kurz E, Hirsch CA, Dalton T, et al. Exercise-induced engagement of the IL-15/IL-15Rα axis promotes anti-tumor immunity in pancreatic cancer. Cancer Cell 2022;40:720-737.e5. [Crossref] [PubMed]
9. Phelps CM, Willis NB, Duan T, et al. Exercise-induced microbiota metabolite enhances CD8 T cell antitumor immunity promoting immunotherapy efficacy. Cell 2025;188:5680-5700.e28. [Crossref] [PubMed]
10. Schmidt T, Jonat W, Wesch D, et al. Influence of physical activity on the immune system in breast cancer patients during chemotherapy. J Cancer Res Clin Oncol 2018;144:579-86. [Crossref] [PubMed]
11. Tomás TC, Eiriz I, Vitorino M, et al. Neutrophile-to-lymphocyte, lymphocyte-to-monocyte, and platelet-to-lymphocyte ratios as prognostic and response biomarkers for resectable locally advanced gastric cancer. World J Gastrointest Oncol 2022;14:1307-23. [Crossref] [PubMed]
12. Choi M, Lee SW, Park W, et al. Can posttreatment blood inflammatory markers predict poor survival in gynecologic cancer?: a systematic review and meta-analysis. Front Immunol 2025;16:1676838. [Crossref] [PubMed]
13. Li Z, An W, Wang B, et al. Prognostic significance of peripheral blood inflammatory biomarkers (SII, PLR, NLR, LMR, SIRI, PIV, PNI) in lip cancer: NLR as an independent biomarker for survival outcomes. Front Oncol 2025;15:1676824. [Crossref] [PubMed]
14. Liu XY, Yao K. Association Between Domain-Specific Physical Activity and Novel Inflammatory Biomarkers Among US Adults: Insights From NHANES 2007-2018. Mediators Inflamm 2025;2025:1989715. [Crossref] [PubMed]
15. CDC National Center for Health Statistics. About NCHS. 2026. Available online: https://www.cdc.gov/nchs/about/index.html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fnchs%2Fabout%2Findex.htm
16. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc 2000;32:S498-504. [Crossref] [PubMed]
17. CDC Physical Activity Basics. Adult Activity: An Overview. 2023. Available online: https://www.cdc.gov/physical-activity-basics/guidelines/adults.html
18. Jeon JY. Exercise as a new therapeutic modality in oncology: CHALLENGE trial refines survivorship care. Nat Rev Clin Oncol 2025;22:900-1. [Crossref] [PubMed]
19. Mak JKL, Tan KCB, Jylhävä J, et al. Walking Speed and Risk of Cancer in Two Prospective Cohort Studies. J Cachexia Sarcopenia Muscle 2025;16:e13792. [Crossref] [PubMed]
20. Fiuza-Luces C, Valenzuela PL, Castillo-García A, et al. Exercise Benefits Meet Cancer Immunosurveillance: Implications for Immunotherapy. Trends Cancer 2021;7:91-3. [Crossref] [PubMed]
21. Cao C, Friedenreich CM, Yang L. Association of Daily Sitting Time and Leisure-Time Physical Activity With Survival Among US Cancer Survivors. JAMA Oncol 2022;8:395-403. [Crossref] [PubMed]
22. Tsukagoshi M, Araki K, Kubo N, et al. Impact of Preoperative Weight Loss on Prognosis in Patients with Pancreatic Cancer. Biomedicines 2025;13:1703. [Crossref] [PubMed]
23. Jia P, Shen F, Zhao Q, et al. Association between C-reactive protein-albumin-lymphocyte index and overall survival in patients with esophageal cancer. Clin Nutr 2025;45:212-22. [Crossref] [PubMed]
24. Liu XY, Zhang X, Zhang Q, et al. The value of CRP-albumin-lymphocyte index (CALLY index) as a prognostic biomarker in patients with non-small cell lung cancer. Support Care Cancer 2023;31:533. [Crossref] [PubMed]
25. Yang M, Lin SQ, Liu XY, et al. Association between C-reactive protein-albumin-lymphocyte (CALLY) index and overall survival in patients with colorectal cancer: From the investigation on nutrition status and clinical outcome of common cancers study. Front Immunol 2023;14:1131496. [Crossref] [PubMed]
26. Zhu D, Lin YD, Yao YZ, et al. Negative association of C-reactive protein-albumin-lymphocyte index (CALLY index) with all-cause and cause-specific mortality in patients with cancer: results from NHANES 1999-2018. BMC Cancer 2024;24:1499. [Crossref] [PubMed]
27. Iida H, Tani M, Komeda K, et al. Superiority of CRP-albumin-lymphocyte index (CALLY index) as a non-invasive prognostic biomarker after hepatectomy for hepatocellular carcinoma. HPB (Oxford) 2022;24:101-15. [Crossref] [PubMed]
28. Wen Y, Zhou Z, Ou Y, et al. Prognostic significance of the CALLY index for cancer risk and survival: evidence from NHANES 2001-2018. World J Surg Oncol 2025;23:431. [Crossref] [PubMed]
29. Gomes-Santos IL, Amoozgar Z, Kumar AS, et al. Exercise Training Improves Tumor Control by Increasing CD8(+) T-cell Infiltration via CXCR3 Signaling and Sensitizes Breast Cancer to Immune Checkpoint Blockade. Cancer Immunol Res 2021;9:765-78. [Crossref] [PubMed]
30. Savage H, Pareek S, Lee J, et al. Aerobic Exercise Alters the Melanoma Microenvironment and Modulates ERK5 S496 Phosphorylation. Cancer Immunol Res 2023;11:1168-83. [Crossref] [PubMed]
31. Pedersen L, Idorn M, Olofsson GH, et al. Voluntary Running Suppresses Tumor Growth through Epinephrine- and IL-6-Dependent NK Cell Mobilization and Redistribution. Cell Metab 2016;23:554-62. [Crossref] [PubMed]
32. Zhu L, Qiao J, Yang C, et al. Prognostic value of lymphocyte-to-monocyte ratio for breast cancer: a systematic review and meta-analysis. BMC Womens Health 2025;25:569. [Crossref] [PubMed]
33. Luo M, Qin L, Li Y, et al. Inflammatory markers from routine blood tests predict survival in multiple myeloma: a Systematic Review and meta-analysis. Front Immunol 2025;16:1669878. [Crossref] [PubMed]
34. Wen X, Sun H, Du S, et al. A nomogram of inflammatory indexes for preoperatively predicting the risk of lymph node metastasis in colorectal cancer. Tech Coloproctol 2024;28:148. [Crossref] [PubMed]
35. Wei Q, Li Z, Feng H, et al. Prognostic value of composite inflammatory prognostic model in pancreatic cancer. Mol Cell Probes 2025;84:102056. [Crossref] [PubMed]

(English Language Editor: J. Gray)