Differential Associations of Cigar, Pipe, and Smokeless Tobacco Use Versus Combustible Cigarette Use With Subclinical Markers of Inflammation, Thrombosis, and Atherosclerosis: The Cross-Cohort Collaboration-Tobacco Working Group.
Jan 27, 2025DOI: 10.1161/circulationaha.124.070852Journal ArticleAHA
Written by: Zhiqi YaoErfan TasdighiZeina A DardariKunal K JhaNgozi OsujiTanuja RajanEllen BoakyeCarlos J RodriguezKunihiro MatsushitaEleanor M SimonsickJoao A C LimaRachel WidomeDebbie L CohenLawrence J AppelAmit KheraMichael E HallSuzanne JuddShelley A ColeRamachandran S VasanEmelia J BenjaminAruni BhatnagarAndrew P DeFilippisMichael J Blaha
Tobacco products are well-known for their highly addictive potential and their association with various adverse health effects.1 According to the 2022 National Survey on Drug Use and Health Annual National Report, there were 50.9 million current tobacco product users in the US.2 In 2021, over 3 million cardiovascular deaths and 7.4 million all-cause deaths were attributable to tobacco use globally.3 Evidence-driven regulatory policies have significantly contributed to the overall decline in cigarette smoking prevalence in recent decades.4,5 However, there has been ongoing development of new non-cigarette tobacco products characterized by rapid initiation and prompt switching use patterns. This poses a challenge in assessing their impact on long-term clinical outcomes.Although there are some studies supporting a potential association between cigar, pipe, and smokeless tobacco use and the development of cardiovascular disease (CVD),6–11 extant research is limited by the relative infrequent use of these products,12 particularly in the National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute (NHLBI) funded cohorts. An advantage of these studies is that they have rich data on markers of inflammation, thrombosis, and subclinical atherosclerosis that are causally linked to and consequently are highly predictive of clinical cardiovascular events.13–15 Studies of subclinical markers related to cardiovascular harm may lend critical insight into the relative toxicity of cigar, pipe, and smokeless tobacco use versus cigarette smoking.16,17,18 These investigations are important to regulatory bodies such as the Food and Drug Administration (FDA) because they provide a reliable assessment of the relative harm of different tobacco products in the absence of clinical event data, and potentially assist in the developing the risk evaluation paradigm and management policy relevant to both current and future tobacco products.To contribute to tobacco regulatory science, we harmonized tobacco-related data and eight commonly measured subclinical markers related to cardiovascular harm from prospective cohorts within the framework of the Cross Cohort Collaboration-Tobacco (CCC-Tobacco) Working Group. Supported by the CCC-Tobacco resource, we aimed to offer estimates regarding non-cigarette tobacco products, including cigar, pipe, and smokeless tobacco use, and their relationship with markers in the domains of inflammation, thrombosis, and subclinical atherosclerosis.
Methods
The data and methods used in this study will be made available exclusively to researchers certified by the CCC-Tobacco and individual cohort committees, only for the purpose of reproducing the results or replicating the procedures.
This study was approved by the Johns Hopkins School of Medicine Institutional Review Board (Approval number: IRB00226738; Date: 24 June 2020). Parent cohorts all had participants sign informed consent and were covered by local IRBs.
In this present study, we leveraged the data from CCC-Tobacco, which is an ongoing, large-scale, cross-continent, collaborative effort involving 23 NIH/NHLBI-funded prospective cohorts to examine tobacco-related cardiovascular health effects.19 The CCC-Tobacco dataset is inclusive of both traditional cardiovascular cohorts (N=13) and non-cardiovascular cohorts (N=10). The rationale of CCC-Tobacco and detailed descriptions of individual cohorts have been described in detail previously.19
The current cross-sectional study selected cohorts that had concurrent data on at least one non-cigarette tobacco product (cigar, pipe, smokeless tobacco) and measurement of at least one of eight a priori-defined subclinical markers, all measured during the same visit. The following 14 cohorts were pooled and harmonized: the Atherosclerosis Risk in Communities (ARIC) Study, the Baltimore Longitudinal Study of Aging (BLSA), the Coronary Artery Risk Development in Young Adults (CARDIA) Study, the Chronic Renal Insufficiency Cohort (CRIC), the Dallas Heart Study (DHS), the Framingham Offspring Study (FHS-OS), the Framingham Heart Study Third Generation (FHS-Gen III), the Jackson Heart Study (JHS), the Health, Aging and Body Composition Study (Health ABC), the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), the Multi-Ethnic Study of Atherosclerosis (MESA), the REasons for Geographic and Racial Differences in Stroke Study (REGARDS), the Rancho Bernardo Study (RBS), and the Strong Heart Study (SHS). The detailed availability of subclinical markers and baseline characteristics across these 14 cohorts are presented in Table S1–S2. The baseline time periods for these cohorts varied, ranging from 1958 to 2011 (Table S1).
After determination of the 14 cohorts, we included participants who had data on 1) combustible cigarette use status; and 2) any of the three non-cigarette tobacco products; and 3) any of the eight subclinical markers measurement, in the harmonized dataset. 98,450 participants met the inclusion criteria.
The use of cigarettes and each of the three non-cigarette tobacco products – cigars, pipes, and smokeless tobacco – was independently valued. All data regarding the use of tobacco products were sourced from comprehensive in-person questionnaires administered by each cohort, and the specific languages they used were summarized in Table S3. Each product was first categorized by use status – never, former, and current. Given the desire to isolate the effects of non-cigarette products, current cigar, pipe, and smokeless tobacco use was further classified into “sole” and “exclusive” use. Sole use was defined as the current use of a given non-cigarette tobacco use without concurrent combustible cigarette use. Exclusive use was defined as current non-cigarette tobacco use, without any history of combustible cigarette use. The conceptual figure depicting the “current”, “sole”, and “exclusive” use of a given non-cigarette tobacco product is shown in Figure S1.
The availability of non-cigarette tobacco use data varied across individual cohorts. Specifically, 13 cohorts had data on cigar use, 11 cohorts had data on pipe use, and 8 cohorts had data on smokeless tobacco use. Within those, data on cigar use were available for 97,564 participants, pipe use for 62,214 participants, and smokeless tobacco use for 66,025 participants. To maximize the sample size of the analysis of each product, combat the potential selection bias, and maintain consistency of the approach across cohorts, we conducted parallel analyses of the three non-cigarette tobacco products. The reference group for cigarette status was never cigarette users. The reference groups for current non-cigarette tobacco users were non-current users of the particular non-cigarette tobacco products, regardless of their history of cigarette use. For example, the reference group of current cigar use was non-current cigar use. For both the sole use and exclusive non-cigarette tobacco use patterns, the reference groups comprised participants who had never used a cigarette or a specific non-cigarette tobacco product.
It is important to note that in this project, the analyses of non-cigarette tobacco products were conducted parallel, and the interpretation of the results should be restricted to the relevant reference group. For example, individuals may be considered “current users” for more than one tobacco product concurrently, reflecting the complexity of poly-tobacco use behaviors in the real world. By analyzing each non-cigarette tobacco product independently with its respective reference group, we are not optimized to conduct formal statistical comparisons of the relative relationships of differing product types with specific biomarkers. As such, direct statistical comparisons across these products should therefore not be considered within the scope of this study.
Subclinical markers were measured according to specified protocols in individual cohorts.19 All markers were meticulously reviewed, cleaned, and harmonized by well-trained CCC-Tobacco staff.
Eight subclinical markers across three domains were assessed in this study:
Inflammation – blood levels of high-sensitivity C-reactive protein (hs-CRP) and interleukin-6 (IL-6).
Thrombosis – blood levels of fibrinogen and D-dimer.
Atherosclerosis — carotid intima-media thickness (cIMT), carotid plaque, coronary artery calcium (CAC), and ankle-brachial index (ABI).
We used best practices to transform inflammation and thrombosis markers to the common units ensuring comparability across cohorts20–23. CCC-Tobacco harmonized blood levels of hs-CRP (N=77,749) in mg/L, IL-6 (N=20,931) in pg/mL, fibrinogen (N=35,900) in mg/dL and D-dimer (N=6,819) in mg/L. However, the variations in the assays across cohorts could lead to modest residual confounding.
cIMT (N=27,803) was measured by a similar protocol across all cohorts using B-mode carotid ultrasound, with some variations in the detailed methods of assessment site and calculation of cIMT across cohorts. For ARIC, we used the mean of maximum values from three segments—left and right common carotid artery (CCA), the bifurcation of the carotid artery, and the left and right internal carotid artery24; CRIC, JHS, CHS, ELSA-Brasil, and MESA measured the cIMT by recording mean of maximum of the wall thickness at the CCA25–29. Carotid plaque (N=24,430) was measured using two-dimensional carotid ultrasound using a Doppler probe and defined as a focal thickening of >1.5 mm within the carotid artery wall.30 When sufficient data were available (N=19,100), we created a harmonized carotid plaque score presented on an integer scale ranging from 0 to 6, reflecting the presence or absence of plaque in the left and right common, bifurcation, and internal carotid arteries. CAC was assessed using a standardized non-contrast cardiac-gated computed tomography scan (N=14,439). All studies quantified CAC using the Agatston score.31 ABI was defined as the ratio of the ankle systolic blood pressure and the brachial systolic blood pressure using standard equipment (N=42,145).32 A resting ABI≤0.9 was considered to reflect subclinical peripheral artery disease.33
The definitions of all demographics, anthropometric, and traditional clinical CVD risk factors have been outlined in the CCC-Tobacco design paper.19 For each key variable, data harmonization adhered to established best practices in the field,19 with this process and all related coding documented in a master file to facilitate replication.
Baseline characteristics were presented across baseline tobacco use status (cigarette: never, former, current; cigar/pipe/smokeless tobacco: current, sole, exclusive) with continuous variables in means (standard deviation (SD)) or medians (interquartile range (IQR)) and categorized variables in percentage (%). The distribution of subclinical markers was presented by current cigarette, cigar, pipe, and smokeless tobacco use groups as well as for the never use group (never use of any tobacco products), to facilitate comparison. The distributions across exclusive non-cigarette tobacco use were also described.
We used multivariable linear and logistic regression models to study the relationship between tobacco product use patterns and subclinical markers. hs-CRP, IL-6, fibrinogen, D-dimer, CAC, cIMT, and carotid plaque were transformed to a natural logarithmic scale because all of these markers were right-skewed. Linear regression models were then applied to the transformed data.34 To address negative infinity, we modeled the CAC score as loge (CAC+1) and the carotid plaque score as loge (carotid plaque score+1). In the linear regression models, all β coefficients were exponentiated and thus should be interpreted as the ratio of geometric means.35 We then presented it as the percentages of higher levels of that marker on the geometric mean scale compared to the reference group.36 Logistic regression models were used to calculate the odds ratio (OR) and study the association between tobacco product use with prevalent CAC>0 (defined as CAC score >0), carotid plaque presence (yes/no), and prevalent ABI≤0.9. In both linear and logistic regression models, the 95% confidence interval (CI) was calculated alongside the point estimates.
All regression models were adjusted for potential confounders including sex, age, race/ethnicity, and individual cohort. Models for current non-cigarette tobacco use were additionally adjusted for smoking status. We encountered no issues with missing data due to the study design and the well-harmonized collection of all relevant covariates in the CCC-Tobacco dataset.
Sensitivity analyses were performed to examine the reliability of the primary results. The body mass index (BMI) is acknowledged as a potential risk factor for numerous subclinical markers, although its influence may not extend uniformly across all such markers. In order to mitigate its potential influence on the relationship between tobacco use and subclinical markers, thereby enhancing the robustness and validity of our findings, we included additional adjustments for BMI in our regression models. BMI was categorized by cutoff points at 25.0, 30.0, and 35.0.37
We also included sensitivity analysis further adjusted for prevalent diabetes and prevalent hypertension, and further adjusted for alcohol use. We repeated our analysis on the sole and exclusive use of non-cigarette tobacco by excluding other types of tobacco use when applicable. For example, in the analysis focused on exclusive cigar use, we excluded participants with any history of cigarette, pipe, or smokeless tobacco use from both the exposure and reference groups.
A two-sided p-value<0.05 indicated statistical significance. STATA (version 18.0, Stata Corp LLC, TX) was used for all analyses.
Results
Table 1 displays the baseline characteristics of the participants by cigarette and non-cigarette tobacco use pattern groups. Among 98,450 participants (mean age of 55.4, SD: 14.1) in this study, 44.8% were men, 29.9% were Black and African American, and 19.2% were Hispanic/Latino individuals. This study included 19,641 current cigarette users, 2,886 current cigar users, 780 current pipe users, and 1,512 current smokeless tobacco users. In general, in comparison to current cigarette users, current cigar and pipe users were similar in age but were more likely to be White men and had higher educational attainment. Smokeless tobacco users were on average older with a balanced racial and ethnic distribution between White and Black individuals, with a more diverse educational background (Table 1).
The distributions of subclinical markers across tobacco use groups are described in Table 2. In general, never users of tobacco had the most favorable profile. The general crude distribution of inflammatory subclinical markers revealed variability in hs-CRP levels among tobacco user and never-user groups, and the lowest mean IL-6 levels were observed among never-users. The crude median and IQR of thrombosis subclinical markers showed less variation across groups. Subclinical atherosclerosis markers showed largely similar distribution by current cigarette, cigar, pipe, and smokeless product use groups. Caution is warranted when interpreting the distributions across exclusive non-cigarette use groups (Table S4) due to the small sample sizes observed in some groups.
Table 3 shows that current cigarette use had strong associations with the blood markers examined in this study, in comparison to never cigarette use. Current smoking was significantly associated with 29% (95% CI: 26%–32%) higher level of hs-CRP and 28% (95% CI: 24%–31%) higher level of IL-6 compared to never smokers in the adjusted linear regression model.
Cigar use exhibited an association with hs-CRP level throughout all current use patterns, with 10% (95% CI: 1%–20%) higher for current cigar use in comparison to non-current cigar use, 19% (95% CI: 12%–26%) higher for sole cigar use, and 11% (95% CI: 1%–21%) higher for exclusive cigar use in comparison to never cigarette use or cigar use. Only sole cigar use was associated with a 15% (95% CI: 6%–24%) higher level of IL-6 in comparison to never cigarette use or cigar use (Table 3).
While current pipe use was not associated with hs-CRP, sole pipe use (23% higher, 95% CI: 8%−40%) and exclusive pipe use (30% higher, 95% CI: 7%–58%) were associated with hs-CRP in comparison to never cigarette use or pipe use (Table 3). For pipe use, the associations with IL-6 were consistently of statistical significance for all current use patterns in relation to their specific reference group (Table 3). For example, IL-6 level was 22% (95% CI: 1%–47%) higher for current pipe use, 32% (95% CI: 10%–57%) for sole pipe use, and 32% (95% CI: 2%–72%) for exclusive pipe use.
There were associations between smokeless tobacco use and hs-CRP from all current use patterns (Table 3). The adjusted model demonstrated a higher level (95% CI) hs-CRP of 11% (2%–20%), 22% (12%–33%), and 16% (3%–31%) for current (vs. non-current smokeless tobacco use), sole, and exclusive smokeless tobacco (vs. never cigarette use and never smokeless tobacco use), respectively. An association between smokeless tobacco users and blood levels of IL-6 was not observed.
Current cigarette use displayed associations with higher levels of fibrinogen (7%, 95% CI: 6%–8%) and D-dimer (9%, 95% CI: 4%–14%) in adjusted linear regression models, in comparison to never cigarette use.
In general, non-cigarette tobacco use was associated with elevated levels of fibrinogen but not associated with D-dimer. Sole cigar use, compared to never cigarette or cigar use, was associated with a 2% (95% CI: 0–4%) higher level of fibrinogen. For pipe, the association was consistently of statistical significance across all patterns, exhibiting 3% (95% CI: 1%–5%) higher fibrinogen for current pipe use (vs. non-current pipe use), 6% (95% CI: 3%–9%) higher for sole pipe use (vs. never cigarette or cigar use), and 4% (95% CI: 0–8%) for exclusive pipe use (vs. never cigarette or cigar use). Similarly for smokeless tobacco use, in the adjusted models, current, sole, and exclusive use demonstrated associations with higher fibrinogen levels by 1% (95% CI: 0–3%), 4% (95% CI: 2%–5%), and 4% (95% CI: 1%–7%), respectively, compared to their specific reference group.
Table 4 shows that current cigarette use was significantly associated with all imaging/physiological markers of subclinical atherosclerosis including CAC, cIMT, carotid plaque, and ABI, in comparison to never cigarette use. Adjusted linear regression models revealed 81% (95% CI: 63%–102%) higher level of CAC, 7% (6%–8%) higher level of cIMT, and 22% (19%–24%) higher level of carotid plaque score was related to current cigarette use in relative to never cigarette use. Current smoking was associated with prevalent CAC>0, prevalent carotid plaque, and prevalent ABI≤0.9 with OR (95% CI) of 1.77 (1.58,1.98), 2.03 (1.88,2.18), and 3.24 (2.88,3.66), respectively.
In comparison to never use of cigarette or cigar, sole cigar use manifested associations with subclinical atherosclerosis markers (Table 4). Specifically, use was linked to a 48% (95% CI: 9%–100%) higher CAC score, an increased likelihood of prevalent CAC > 0 with an OR (95% CI) of 1.49 (1.06–2.10), a 3% (95% CI: 1%–5%) increase in cIMT, a 9% (95% CI: 3%–15%) higher carotid plaque score, and a higher prevalence of carotid plaque with an OR (95% CI) of 1.38 (1.08–1.75). However, neither current cigar use nor exclusive cigar use was associated with any of the subclinical atherosclerosis markers.
While current pipe use and exclusive pipe use did not display a relationship with the subclinical atherosclerosis measures, there were associations between sole pipe use and carotid plaque-related markers, with 4% (95% CI: 1%–7%) higher level of cIMT, 10% (95% CI: 4%–17%) higher level carotid plaque score, and higher risk of prevalent carotid plaque with OR (95% CI) of 1.43 (1.08–1.89), compared to individuals who did not use either cigarettes or pipes (Table 4).
The use of smokeless tobacco was associated with select subclinical atherosclerosis markers (Table 4). In the adjusted linear regression models, current (vs. non-current use of smokeless tobacco) and sole smokeless tobacco use (vs. never use of cigarette or smokeless tobacco) were associated with 4% (95% CI: 2%–6%) and 7% (95% CI: 4%–9%) higher levels of cIMT, respectively. Associations were observed between current, sole, and exclusive (vs. never use of cigarette or smokeless tobacco) smokeless tobacco use and carotid plaque score, with 9% (95% CI: 3%–14%), 18% (95% CI: 13%–24%), and 11% (95% CI: 5%–19%) higher, respectively. Additionally, current, sole, and exclusive smokeless tobacco use were associated with prevalent carotid plaque with OR (95% CI) of 1.31 (1.11–1.55), 1.68 (1.38–2.04), and 1.43 (1.08–1.87), respectively.
Similar to previous publications, the present study computed subclinical marker sensitivity as the marker exhibiting the greatest responses to specific tobacco product exposure.38–40 We empirically quantified higher levels of specific subclinical markers and considered more than 25% as strong sensitivity, 6–24% as moderate sensitivity, and less than 5% as mild sensitivity. Considering all tobacco products, hs-CRP merged as the most sensitive marker (10%, 19%, 11% higher with current, sole, and exclusive cigar use, 23% and 30% higher with sole and exclusive pipe use, 11%, 22%, and 16% higher with current, sole, and exclusive smokeless tobacco use, respectively), followed by IL-6, which was particularly sensitive to variations in pipe use patterns. Besides these, the carotid plaque score also exhibited higher levels with largely moderate sensitivity among non-cigarette users. Although fibrinogen and cIMT also displayed statistically higher levels across non-cigarette use, the percent of higher levels were mainly less than 5% and thus were considered as overall mild sensitivity. Figure 1 depicts the percent higher levels in the two most sensitive subclinical markers, hs-CRP and IL-6, associated with cigar, pipe, and smokeless tobacco use compared to specific tobacco never users.
Overall, the sensitivity analyses generated similar results to the main analyses. Upon introducing additional adjustments for BMI as a categorized variable, the results remained consistent with the primary analyses (Tables S5–S6). When further adjusting for prevalent diabetes and hypertension (Tables S7–S8), and alcohol use (Tables S9–S10), we observed largely similar point estimates and directionally identical results. When the analyses were repeated by excluding participants with other types of tobacco use when data was available, point estimates were largely similar, and results were directionally the same as our main results (Table S11–S12).
Discussion
Herein, we have described the largest study to date, spanning a broad time range from 1958 to 2011, to assess both cigarette use and non-cigarette tobacco use and their association with subclinical markers related to subclinical cardiovascular harm across three domains – inflammation, thrombosis, and atherosclerosis. Cigarette use was consistently associated with all cardiovascular subclinical markers assessed in this study. The non-cigarette tobacco products considered were also associated with higher circulating concentrations of inflammatory subclinical markers including hs-CRP and IL-6, with moderately higher levels of cIMT and carotid plaque score, and mildly higher blood level of fibrinogen. Taken together, these results provide insight into the potential cardiovascular harm effects associated with cigars, pipes, and smokeless tobacco use. To note, our findings regarding non-cigarette tobacco use are specific to data from U.S. populations during the studied time frame and may not reflect the effects of contemporary products or products characteristic of use in other countries (i.e. Snus in Sweden).
Aligned with the present results, several smaller studies have also demonstrated the associations between the use of non-cigarette tobacco products and elevated levels of inflammatory subclinical markers,41–44 as well as thrombosis subclinical markers.45 For instance, findings from the Caerphilly Prospective Heart Disease Study revealed a 22.8% increase in CRP associated with pipe/cigar use.41 The study conducted by Muddathir et al. highlighted a significant increase in fibrinogen with pipe use.45 Our findings with tobacco products corroborate these findings and support the observation that the use of non-cigarette tobacco products is associated with significant cardiovascular risk. Importantly, our results also show for the first time that several product use patterns are associated with indices of subclinical disease, which further attests to the deleterious impact of tobacco products. However, our findings differ from those of the Population Assessment of Tobacco and Health (PATH) Study Wave 1, which found no associations between smokeless tobacco products and elevated markers of inflammation or thrombosis, in comparison to never users.46–47 Our findings reflect general smokeless tobacco use in the U.S. over a broad time span, which does not specifically include modern products. The participants in our study were from a different era who have different smokeless tobacco use habits from the PATH Study, and we also examined a few different subclinical biomarkers. Further, different from our design, the PATH analysis excluded sole smokeless tobacco users and involved a relatively young study population with a smaller sample size than ours. In contrast, our study was conducted using a well-harmonized dataset with a large sample size, enabling us to capture the statistically significant relationships between smokeless tobacco use and markers of inflammation and thrombosis. Further, in contrast to the PATH Study, we have observed notable results in both sole and exclusive smokeless tobacco use.
In our findings, sole use of non-cigarette tobacco products showed stronger cross-sectional associations with subclinical atherosclerosis markers compared to current or exclusive use patterns for each type of product. Unlike inflammation and thrombosis markers, which respond more immediately to harmful substances, atherosclerosis markers reflect longer-term cumulative effects and thus may somewhat capture the cumulative impact of prior smoking. Therefore, sole users in our study, who may have a history of cigarette use, demonstrate stronger associations with subclinical atherosclerosis markers compared to exclusive users. The interpretations of the findings of sole use across non-cigarette tobacco should focus on patterns rather than inferring causation.
Evaluating specific subclinical markers associated with tobacco use significantly enhances our understanding of the health risks associated with tobacco products.38,48–51 Subclinical markers are particularly valuable in the case of tobacco products, where their influence may manifest with a latency period, serving as useful indicators to assess risks well in advance of clinical events. Our findings from a large, harmonized dataset covering three distinct domains of subclinical cardiovascular harm introduce a new perspective and an epidemiologic analytical framework that is useful for assessing the relative risk of different tobacco products. Importantly, data derived from concurrent measurements of the effects of tobacco products across different domains of cardiovascular harm are useful in assessing the relative sensitivity of different biological processes to tobacco products.
In view of the high interest of the FDA and other regulatory agencies in scientific evidence regarding the health effects of tobacco products,52,53 we believe that the findings of our study have distinct value for tobacco control and regulation. First, our findings underscore the consideration of a risk continuum when regulating different tobacco products. By delineating the associations between various non-cigarette tobacco use patterns and specific subclinical markers, the present study lends insight into the distinct risks of each product, which could inform the development of formal risk determinations. This understanding may facilitate a more tailored regulatory approach, ensuring that regulations align with the relative risks associated with each product. Second, as the tobacco landscape evolves with the introduction of new and potentially related products, the findings of this study offer guidance on future product evaluation and effective regulatory strategies to ensure that products introduced in the market do not impose greater risk. The identification of hs-CRP and IL-6 as the most sensitive subclinical markers with consistent associations across non-cigarette tobacco use patterns suggests their potential utility in assessing the relative health impact of emerging products. This insight may inform future study design, especially when evaluating data within applications for potential modified risk tobacco product status.
There are several strengths of our study. First, by consolidating multiple high-quality cohorts, the sample size of CCC-Tobacco enhanced the statistical power facilitating a comprehensive evaluation of the associations between non-cigarette tobacco products and subclinical markers and allowing for a direct comparison of effect sizes with cigarettes.19 Second, owing to the innate designs and heterogeneous characteristics of the individual cohorts, our study offers high external validity to the general population. Third, this study is the first to illustrate a comprehensive mapping of commonly assessed subclinical markers indicative of inflammation, thrombosis, and atherosclerosis in population-based studies, providing a template study of emerging tobacco products.
This study also has limitations. Our study did not have measures of intensity or cumulative pack-year of non-cigarette tobacco product use, representing a gap in our understanding of the health impact of tobacco products. Individual cohorts were established across an extended timeframe, allowing potential changes in the identity and content of tobacco products over time that could not easily be accounted for.54 The dual or poly use patterns were not explored in this study due to the limited sample size of specific clusters. We conducted analyses with multiple exposures and nine subclinical measures, and we did not account for multiple statistical testing. In addition, although some of the findings were statistically significant, some of the associations were of modest magnitude and may not be clinically substantive. As this is a population-level study to support regulatory science, we did not adjust for factors such as diabetes, hypertension or alcohol use as one would in a pathophysiology study or a risk prediction project (although when checked in exploratory analyses, these did not meaningfully impact our results). Instead, we focused on adjusting for demographic and socioeconomic factors to ensure the generalizability and applicability of our findings at the population level, within the context that these products are commonly used. We were unable to account for the exact time of tobacco use prior to the blood draw due to the study design of the individual cohorts however nearly all participants were instructed to fully fast for 8 hours before the baseline exam. We adjusted for individual cohorts to account for potential variations in study protocols, and as such we believe our study accurately reflects chronic background tobacco use but may not be used to reflect the effects of acute impacts tobacco use. Also, the cross-sectional and observational study design limits our ability to establish causal associations. The conclusions may be limited in their generalizability to Asian and Alaska Native/American Indian populations, as these groups were relatively underrepresented in some of our analyses. Finally, there remains the possibility of residual confounding, perhaps attributable to unmeasured socioeconomic determinants.
Conclusions
The findings of our study revealed that the use of non-cigarette tobacco products was associated with subclinical cardiovascular harm, as indicated by relevant inflammation, thrombosis, and atherosclerosis markers. Sensitive subclinical markers, such as hs-CRP and IL-6, could be valuable in regulatory science to assess the harmful health effects of short-term non-cigarette tobacco use before clinical cardiovascular disease is apparent, given their consistently elevated levels among users of these products.
AK
Dr. Amit Khera
Cardiology · Preventive CardiologyThe University of Texas Southwestern Medical Centeramitk
