Vigorous Intermittent Lifestyle Physical Activity and Cancer Incidence Among Nonexercising Adults

1The association between physical activity (PA) intensity and certain cancer sites, such as breast and colon cancers, is dose dependent and has a greater risk reduction associated with vigorous physical activity (VPA) compared with lower intensities. Although VPA is time efficient, structured exercise bouts may not be feasible or appealing to most middle-aged adults. Vigorous intermittent lifestyle physical activity (VILPA) refers to brief and sporadic (eg, up to 1-2 minutes [min]) bouts of VPA during daily living, eg, bursts of very fast walking or stair climbing. Therefore, VILPA cannot be captured by questionnaires; wearable trackers are essential. A recent study found a beneficial association of daily VILPA with total cancer mortality, although the low number of cancer deaths precluded a detailed dose-response examination. To our knowledge, no study has evaluated the associations of VILPA with cancer incidence. In a large sample of inactive adults (nonexercisers) we assessed the dose-response associations of device-measured daily VILPA duration with incidence of cancer and estimated the minimum VILPA amounts for cancer risk reduction.

2Details of the sample selection and study methods are described in eFigure 1 and the eAppendix in Supplement 1. We used the UK Biobank wrist accelerometry substudy activity data, a study previously approved by the UK National Research Ethics Service (No. 11/NW/0382) that had obtained written informed consent from all participants. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (eTable 5 in Supplement 1).

3From the UK Biobank wrist accelerometry substudy activity data, we included only nonexercisers—participants who reported no leisure time exercise and 1 or fewer recreational walks per week (eTable 1 in Supplement 1). We excluded participants with missing covariates; prevalent cancer or previous cancer in remission; cancer event during the first year after accelerometry baseline; or inadequate wear time (eTable 2 in Supplement 1). Physical activity intensity was classified using a validated, 2-stage machine learning-based Random Forest activity classifier covering VPA, moderate-intensity physical activity (MPA), and light-intensity physical activity (LPA; eAppendix in Supplement 1). We tested bouts of up to 1 or 2 min, based on recent data showing that the mean (SD) time required to reach vigorous intensity during 5 typical VILPA activities is 73.5 (26.2) seconds. Cancer incidence was defined as cancer registration, hospitalization for cancer, or death attributed to any cancer, and excluded in situ, nonmelanoma skin cancer, and non−well-defined cancers. We also derived a composite cancer outcome of 13 sites that have been shown to be associated with PA (eTable 3 in Supplement 1). Participants were followed up through October 30, 2021 (mortality and hospitalizations), or June 30, 2021 (cancer registrations).

4We assessed dose response of adjusted absolute risk between daily VILPA duration and incident cancer using Poisson regression. Time-to-event associations of daily VILPA duration were analyzed using Fine-Gray subdistribution hazards that account for competing risks from non−outcome-related causes of death. Analyses were adjusted for age, sex, body mass index (calculated as weight in kilograms divided by height in meters squared), education level, smoking status, alcohol consumption, sleep duration, fruit and vegetable consumption, medications, parental cancer history, prevalent cardiovascular disease, daily durations of all LPA and MPA, and daily duration of longer VPA bouts, as appropriate (eTable 4 in Supplement 1). We estimated, the minimal VILPA dose as that associated with 50% of the optimal risk reduction. We used evenly spaced knots at the 6th, 34th, and 67th percentiles to fit the right-skewed VILPA distribution; E-values estimated the plausibility of unmeasured confounding., The interpretation of data was based on 95% CIs across the dose response curves (not P value testing). Data analyses were performed February to March 2023 using R, version 4.2.1 (The R Foundation for Statistical Computing) with RMS, version 6.3.0 (Harrell FE) and Survival package, version 3.3.1 (RStudio).

5The study sample comprised 22 398 participants (mean [SD] age, 62.0 [7.6] years; 10 122 [45.2%] men and 12 276 [54.8%] women; 303 [1.4%] Asian, 237 [1.1%] Black, 142 [0.6%] multiracial, 21 509 [96.0%] White, and 207 [0.9%] individuals of other race/ethnicity; eTable 6 in Supplement 1). During a mean (SD) follow-up of 6.7 (1.2) years (149 650 person-years), 2356 new cancer events occurred (1084 in PA-related cancer sites); 6.2% of the study participants recorded no VILPA.

6Most (92.3%) VILPA was accrued in bouts of up to 1 min while 97.3% of all bouts lasted up to 2 min. For both duration lengths, the median daily average VILPA was 4.5 min and the maximum was 16.0 min. Adjusted dose-response curves of absolute risk (eFigure 2 in Supplement 1; Figure 1) and relative hazard ratios (HR) indicated a near-linear association of both VILPA bout lengths with total and PA-related cancer incidence (Figures 2 and 3). The dose-response curves were steeper, and the magnitude of the risk reduction sharper for PA-related cancer than for total cancer incidence. For example, the minimum dose,, for VILPA bouts of up to 1 minute was 3.4 (total cancer) min/d and 3.7 (PA-related cancer) min/d (HRs, 0.83; 95% CI, 0.73-0.93; and HR, 0.72; 95% CIs, 0.59-0.88, respectively) (eTable 7 and eFigures 13 and 14 in Supplement 1). The daily median VILPA duration (4.5 min/d) was associated with an HR of 0.80 (95% CI, 0.69-0.92) and 0.69 (95% CI, 0.55-0.86).

7Results for VILPA bouts of up to 2 min were very similar. In the sensitivity analyses, excluding participants who were underweight or in poor health or had an event in the first 2 years (eFigure 3 in Supplement 1) and removing BMI from the models (eFigure 4 in Supplement 1) did not appreciably change the results. For the estimates to be attenuated to null, the association of an unmeasured confounder with exposures and incident cancer would need to have a HR of 1.70 (95% CI lower bound, 1.36) and 2.12 (95% CI lower bound, 1.53) for PA-related cancer (eTable 8 in Supplement 1). Additional adjustment for VILPA energy expenditure (kilojoules/kg/d) had minimal effect on the dose-response curves (eFigure 5 in Supplement 1). eFigure 6 in Supplement 1 presents the nested models sequentially adjusted for groups of confounders, and eFigure 7 in Supplement 1 presents a categorical VILPA exposure analysis. Adjustment for smoking pack-years revealed a modest attenuation of the associations for PA-related cancer (eFigures 8 and 9 in Supplement 1). Adjustments for detailed alcohol consumption (eFigures 10 and 11 in Supplement 1) and for prevalent diabetes (eFigures 12 and 13 in Supplement 1) showed minimal associations.

8Nonexercising adults, the majority of the middle-aged population,, are at an increased risk of developing certain cancers. We found inverse associations of modest VILPA amounts with total cancer and, in a more pronounced manner, PA-related cancer incidence. Although steeper risk reductions occurred at the lower end of the VILPA distribution (up to approximately 4-5 min/d), there were continuing gains with higher daily VILPA amounts. With little variation between bouts of up to 1 or 2 min, a minimum of 3.4 to 3.6 min of VILPA/d was associated with a 17% to 18% reduction in total incident cancer risk (compared with no VILPA). The study sample median of 4.5 VILPA min/d was associated with a 31% to 32% reduction in PA-related cancer incidence. For comparison, 1 metabolic equivalent unit increase in cardiorespiratory fitness (3.5 mL of oxygen uptake/kg/min) is associated with a 7% reduction in total cancer risk.

9Proof-of-concept trials have shown that small doses of intermittent VPA may produce rapid improvements in cardiorespiratory fitness, providing a potential biological explanation of the associations with incident cancer mortality observed in the present study findings. More VPA has also been specifically associated with reduced risk of breast, endometrial, and colon cancers. The main biologic pathways associating PA and cancer incidence are inflammation, insulin resistance, body composition, and endogenous sex hormones. Although the present and previous studies were observational and could not confirm causation, a recent Mendelian randomization analysis provided evidence of a causal association between VPA and breast cancer. To our knowledge, this study is among the first to use a wrist accelerometry classifier to estimate VILPA.,

10A limitation of this study was that responses to the leisure-time exercise questions, which determined the sample inclusion criteria, were administered an average of 5.5 years before the accelerometry baseline was recorded. However, the questions had high stability over time (88%) among the 6095 participants who had repeated examinations.

11This cohort study found that daily VILPA duration was inversely associated with incident cancer risk in a near-linear manner, with steeper dose-response for PA-related cancers. As few as 4 to 5 min of VILPA daily was associated with a substantially lower cancer risk. Long-term trials with cancer-related biomarker outcomes and well-designed cohort studies with wearable devices should further explore the potential of VILPA as a cancer prevention intervention for nonexercising individuals and for those who find structured exercise unappealing.