Association of Depressive Symptoms with Cancer Risk in Middle-Aged and Older Chinese Adults: A Longitudinal Cohort Study

Aims. Although, the association between depression and cancer has already been very well studied, but studies about depressive symptoms and cancer development are scarce and inconclusive among Chinese adults. We aimed to investigate whether the severity of depressive symptoms is associated with total and site-specific cancer risk in a nationally representative sample of middle-aged and older Chinese adults. Methods. This study was based on the China Health and Retirement Longitudinal Study (CHARLS), which was initiated in 2011 and followed up to 2018. We included 11,974 individuals aged 45 years or older with complete information about depressive symptoms and no history of cancer at baseline. Depressive symptoms were assessed using the validated 10-item of Center for Epidemiologic Studies Depression Scale (CES-D). Incident cancer cases were documented in the biennial self-reported questionnaires. The multivariable Cox proportional hazards model was used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs). Results. Of 11,794 individuals included in the analysis, 51.72% were women. After a median follow-up of 7.0 years, 265 incident cancer cases were identified. Overall, there was a significant positive association between CES-D score and cancer risk (HR 1.02, 95%CI 1.00-1.05, P=0.03). Severe depressive symptoms were associated with a 75% increased cancer risk (HR 1.75, 95%CI 1.10-2.78). Such associations were evident among women rather than men. In the cancer-specific analysis, the association of severe depressive symptoms and cancer risk was more pronounced for female hormone-related cancers (HR 5.58; 95% CI 2.70-11.54). Conclusions. The findings imply that individuals with severe depressive symptoms could be considered as high-risk population in cancer screening programs.

Validation of Self-Reported Cancer Diagnosis by Cognitive Status in the Health and Retirement Study

Exploring the relationship between cognition and cancer is increasingly important as the number of older adults in the US grows. The Health and Retirement Study (HRS) has longitudinal data on cognitive status and self-reported cancer diagnoses, but these self-reports have not been validated. Using HRS linked to Medicare Fee for Service (FFS) claims (1998-2016), we evaluated the validity of self-reported cancer diagnoses (excluding non-melanoma skin) against Medicare claims by respondent cognitive status. We included 8,280 Medicare-eligible HRS participants aged ≥67 with at least 90% FFS coverage. Cognitive status was ascertained from the HRS interview following the date of cancer diagnosis (or reference claim date) using the Langa-Weir method and was classified as normal, cognitive impairment no dementia (CIND), or dementia. We calculated the sensitivity, specificity, and Cohen's kappa for first incident malignant cancer diagnosis by cognitive status group. The majority (76.4%) of participants scored as cognitively normal, 9.6% had CIND, 14.0% had dementia and, overall, 1,478 had an incident cancer diagnosis. Among participants with normal cognition, sensitivity of self-reported cancer diagnosis was 70.2% and specificity was 99.8% (kappa=0.79). Among participants with CIND, sensitivity was 56.7% and specificity was 99.8% (kappa=0.66). Among participants with dementia, sensitivity was 53.0% and specificity was 99.6% (kappa=0.64). Results indicate poor validity of self-reported cancer diagnoses for older adults with CIND or dementia. These findings suggest researchers interested in cancer and cognition should use the HRS-Medicare linkage to ascertain cancer diagnosis from claims, and they highlight the importance of cognitive status in research among older adults.

Memory Aging Phenotypes Among Older Cancer Survivors: A Latent Growth Analysis of the Health and Retirement Study

While cancer survivors experience many long-term health effects, there is limited evidence on the potentially heterogeneous memory aging of older cancer survivors. We identified memory aging phenotypes of older US cancer survivors, and determined sociodemographic and health-related predictors of membership. Data were from 2,755 survivors aged ≥50 in the U.S. Health and Retirement Study (1998 – 2016). Self-reported first incident cancer diagnosis (except non-melanoma skin cancer) and memory (composite immediate and delayed word-list recall score, combined with proxy-reported cognition) were assessed at biennial interviews. Memory aging phenotypes were identified using latent growth curve (LGC) models, with baseline being time of cancer diagnosis. Logistic regression evaluated predictors of group membership. 5 distinct memory aging groups were identified: low memory (n=165, 6.16%); medium-low memory (n=459, 17.1%); medium-high memory (n=733, 27.4%); high memory (n=750, 28.0%); and very high memory (n=571, 21.3%). The low memory group received less chemotherapy compared to the other groups (20.0% vs. 25.5%, 31.7%, 36.8%, 41.5%%, respectively), and had the shortest mean survival time after diagnosis (1.08 vs 2.10, 2.76, 3.37, 4.31 years, respectively). Older age at diagnosis (OR: 1.71, 95%CI: 1.61-1.82), being male (OR: 4.10, 95%CI: 2.82-6.51), having a history of stroke (OR: 4.62, 95%CI: 2.57-8.30) and depression prior to diagnosis (OR: 1.19, 95%CI: 1.05-1.34) were independently associated with being in the low memory group vs. the medium-high memory group. We identified distinct memory aging phenotypes among older cancer survivors. Further research should evaluate the influence of pre-cancer memory and how these phenotypes differ from the general population.

506 Role of inflammation and comorbidities in the association between heart failure and incident cancer in the HUNT3 cohort

Aims Conflicting data exist regarding the risk of cancer in patients with heart failure (HF). It was first reported that incident cancer is more common among patients with than without HF, whereas more recent studies indicate that this association is primarily driven by comorbidities. HF, cancer, and comorbidities, such as chronic kidney disease (CKD) and chronic obstructive pulmonary disease (COPD), share numerous risk factors, including a state of chronic low-grade inflammation reflected by elevated circulating levels of pro-inflammatory cytokines. The objective of this analysis was to assess whether chronic low-grade inflammation, as measured by levels of high-sensitivity C-reactive protein (hsCRP), and comorbidities mediate the association of HF with incident cancer. Methods We used data from the 3rd wave of the Nord-Trøndelag Health Study (HUNT3), a population-based study that enrolled 50 803 individuals ≥18-year-old between October 2006 and June 2008 in the Nord-Trøndelag County (Norway), and from the administrative health care records of the same region. Associations between baseline characteristics and the development of cancer were assessed using Cox proportional hazards regression models, using time from HUNT3 enrolment as the time scale. Analyses were performed using R statistical software, version 4.0.2. Results In HUNT3, hsCRP was measured in 47 571 individuals at the time of enrolment. Of these, we excluded 2308 patients because of missing information, leaving a cohort of 45 263 subjects. The prevalence of cardiovascular disease, comorbidities, and obesity was progressively higher with increasing concentrations of hsCRP. During a median follow-up of 12 years, there were 66/408 cases of incident cancer in patients with HF at baseline and 5024/47 163 in subjects without HF, with a more than 2-fold (HR 2.30; 95% CI 1.80–2.93; P < 0.001) increase in risk of developing cancer. After adjusting for age and sex, the excess risk decreased to 43% (HR 1.43; 95% CI 1.12–1.82). When including hsCRP in the model, the HF-related risk of cancer was 33% (HR 1.33; 95% CI 1.04–1.70; P = 0.022). Furthermore, when body mass index, CKD, COPD, and smoking and drinking habits were included in the model, the risk of cancer in HF patients compared to individuals without HF was no longer significant (HR 1.23; 95% CI 0.94–1.60; P = 0.127). Age, male sex, hsCRP, COPD, obesity, and smoking habits were all associated with an increased risk of cancer. Conclusions The increased risk of cancer in HF patients compared with the general population is at least in part explained by concomitant inflammation and comorbidities.

Demographic and Epidemiological Contributions to Recent Trends in Cancer Incidence in Hong Kong

Background: Hong Kong has an ageing Chinese population with high life expectancy and a rising number of cancer cases. While population ageing could lead to higher incidence, we aim to quantify the demographic and epidemiological contributions to this trend by disentangling the effect of these factors. Methods: We analysed secular trends of cancer incidences of all cancer sites combined, including the five top cancers in men and women in Hong Kong in 1983–2017, by disentangling effects of demographics (ageing population and population growth) and cancer risk/rate change using the RiskDiff methodology. Results: Overall, age-standardised incidences of all cancers combined in women and in men declined over the study period (−5.3% for women, −30.2% for men), but total incident cancer case counts increased dramatically (156.5% for women, 96% for men). This increase was primarily due to ageing and increasing population (95% age, 66.1% growth for women, and 119.4% age, 25.4% growth for men), while disease risk for all cancers combined has a decreasing trend (−4.5% for women and −48.8% for men). For the site-specific risk changes among the most five common cancer types, there were increases in risks of prostate and colorectal cancers in men, and breast, endometrial, and thyroid cancers in women. Conclusion: Demographic changes and ageing in our Chinese population resulted in a marked increase in the number of cancer diagnoses in Hong Kong in past decades. The surge in incident case counts overall is expected to stress the healthcare system in terms of the increased demand of healthcare professionals. Cancer surveillance should be enhanced in view of the growing demand from older patients and the cancer types with fast-increasing incidence rates in our population.

Changes in life expectancy and disability-free life expectancy in successive birth cohorts of older cancer survivors: a longitudinal modeling analysis of the US Health and Retirement Study

The population of older cancer survivors in the US is rapidly growing. However, little is currently known about how the health of older cancer survivors has changed over time and across successive birth cohorts. Using data from the US Health and Retirement Study, we parameterized a demographic microsimulation model to compare partial cohort life expectancy (LE) and disability-free LE for US men and women without cancer and with prevalent and incident cancer diagnoses for four successive 10-year birth cohorts born 1918-1927 to 1948-1957. Disability was defined as being disabled in ≥1 activity of daily living. These cohorts had mid-point ages of 55-64, 65-74, and 75-84 years during the periods 1998-2008 (the “early” period) and 2008-2018 (the “later” period). Across all cohorts and periods, those with incident cancer had the lowest LE, followed by those with prevalent cancer and cancer-free individuals. We observed declines in partial LE and an expansion of life spent disabled among more recent birth cohorts of prevalent cancer survivors. Our findings suggest that advances in treatments that prolong life for individual cancer patients may have led to population-level declines in conditional LE and disability-free LE across successive cohorts of older cancer survivors.

Validation of self-reported incident cancer diagnoses in the U.S. Health and Retirement Study: A tool for population-based cancer and aging research.

312 Background: Population aging and improving cancer survival rates are resulting in a growing population of older cancer survivors in the United States (US). As a result, there is an increasing need for longitudinal, population-representative data for interdisciplinary cancer research among older adults. The US Health Retirement Study (HRS) is an ongoing population-representative cohort of US adults over age 50 that contains rich interview and biomarker data on health during aging. Interviews have collected self-reported cancer diagnoses since 1998, but these self-reports have not been validated. We compared first incident cancer diagnoses self-reported in HRS interviews against diagnostic claims from linked Medicare records. Methods: We examined the validity of first incident cancer diagnoses self-reported in biennial HRS interviews from 2000 through 2016 against ICD-9 and ICD-10 diagnostic claim records among 8,242 HRS participants aged ≥65 with 90% continuous enrollment in fee-for-service Medicare, using the claim records as the gold standard. We calculated the sensitivity, specificity, and k for first incident cancer diagnoses (all cancers combined, excluding non-melanoma skin cancer, and each of bladder, breast, colorectal/anal, uterine, kidney/renal, lung/bronchus, and prostate cancers) cumulatively over the follow-up, and at each biennial study interview. Results: Self-reports of first incident cancer diagnosis (agnostic of site) between 2000 and 2016 had 73.2% sensitivity and 96.2% specificity against Medicare claims (k = 0.73). For site-specific self-reports, sensitivities ranged from 44.7% (kidney) to 75.0% (breast), and specificities ranged from 99.2% (prostate) to 99.9% (bladder, uterine, and kidney). Results were similar in sensitivity analyses restricting to individuals with 100% continuous fee-for-service Medicare enrollment and when restricting to individuals with at least 24 months of Medicare enrollment. Conclusions: Self-reported cancer diagnoses in the HRS have reasonable validity for population-based research on cancer and aging across cancer types. Apart from breast cancer, cancer site specific analyses will greatly benefit from the improved validity of self-report with Medicare claim linkage.

Impact of adverse events on quality of life and hospital costs in secondary cardiovascular disease prevention

Background and purpose People with atherosclerotic vascular disease remain at high risk of cardiovascular (CVD) events despite effective risk factor management 1. There is little research on impacts of adverse events on quality of life (QoL) and hospital cost to inform evaluations of novel interventions in this population. We estimate QoL and annual hospital costs associated with a range of adverse events of interests using the individual participant data from the Randomized Evaluation of the Effects of Anacetrapib through Lipid Modification (REVEAL) trial. Methods Data from the 30,449 participants with atherosclerotic vascular disease receiving effective statin therapy in REVEAL, were used to estimate regression models for participants' hospital costs and QoL using participants' characteristics at entry (socio-demographic, clinical, prior diseases and treatments) and time-updated adverse events. We estimate costs and QoL in the year of an event, and in subsequent years, using stepwise covariate selection (p-value <0.01). Standard errors were adjusted for clustering of participant annual costs. Hospital episodes were costed (2019 UK£) using the UK Healthcare Resource Groups reference costs 2. One- and two-part generalized linear regression models (GLMs) for annual hospital costs (part 1: logistic model for estimating probability of incurring cost, part 2: GLM with Gaussian, Poisson or Gamma distributions with identity or log links for estimating costs, conditional on incurring any) were compared. EQ-5D-5L questionnaires, completed by study participants at entry and final follow-up visits in the study, were mapped into QoL utility scores 3. QoL utility at final follow-up was used to estimate QoL decrements of adverse events using GLM linear model and adjusting for QoL at entry in addition to other participants characteristics. Results The two-part model with gamma distribution and identity link, indicated by specification tests and model fit statistics, was selected for modelling annual hospital costs (Figure 1). Non-haemorrhagic stroke, non-coronary revascularization, coronary revascularization and incident cancer were associated with highest hospital costs. The QoL model (Figure 2) indicated large QoL decrements associated with non-fatal non-haemorrhagic stroke, heart failure hospitalization, incident cancer and non-coronary revascularization, and comparatively small QoL decrement associated with experiencing non-fatal myocardial infarction. Conclusion These cost and QoL models in a well-managed contemporary high CVD risk patient population would assist in assessments of long-term net effects and cost-effectiveness of novel interventions to reduce cardiovascular risk. FUNDunding Acknowledgement Type of funding sources: Other. Main funding source(s): Merck Sharp & Dohme and UK Medical Research Council Figure 1 Figure 2