Introduction

Nonalcoholic fatty liver disease (NAFLD) is rapidly becoming the most common chronic liver disease, with a global prevalence of 38% [1]. NAFLD often precedes type 2 diabetes mellitus (T2DM) [2], which is one of the most significant extrahepatic complications [3, 4]. Considering that there are no approved pharmacological treatments, lifestyle modification is necessary but challenging to reduce the risk of T2DM in patients with NAFLD [5].

Cigarette smoking has a significant negative impact on public health, causing over 480,000 deaths each year [6, 7]. Smoking has been reported as a risk factor for NAFLD and may accelerate the progression of liver disease [8,9,10]. Therefore, smoking cessation is recommended for NAFLD patients [11, 12]. However, smoking cessation can be complicated by weight gain [13], which typically occurs within 5–7 years after quitting smoking [14, 15]. Post-cessation weight gain poses a potential health concern, especially in NAFLD patients [16]. Weight gain is an important risk factor for the development of T2DM in NAFLD patients [17, 18]. To date, clinical evidence about the impact of post-cessation weight gain on NAFLD and its comorbidities remains limited [19]. Furthermore, post-cessation weight gain limits the willingness of patients and reduces the success of cessation attempts. It has been reported to be a significant cause of relapse [20]. Therefore, it is important to assess the impact of weight change after smoking cessation on patients with NAFLD. Appropriate management of post-cessation weight could maximize its health benefits.

In this large-scale cohort study, we aimed to assess the effects of smoking cessation and subsequent weight change on the risk of incident T2DM in NAFLD and NAFLD-free smokers. These findings will facilitate the provision of smoking cessation advice to patients with NAFLD in clinical practice.

Methods

Study population

The study population was composed of adults aged ≥ 18 years who attended annual health checkups at Zhenhai Lianhua Hospital in 2007. Due to the low proportion of female smokers in China, we included only males in this study. We excluded the following participants: (i) female subjects (ii) those with missing information on hepatic ultrasound, smoking status, or body weight and (iii) those with excess alcohol intake or a history of chronic liver disease at baseline, including viral hepatitis (such as hepatitis B and hepatitis C), drug-induced liver disease, autoimmune liver disease, and genetic metabolic liver diseases (such as Wilson's disease). During the follow-up, we further excluded the following participants: (i) those who were diagnosed with T2DM at baseline and (ii) those who were lost to follow-up. After the exclusions, a total of 6,559 participants were included for follow-up until December 2014 (Supplementary Figure S1). Previous studies have shown that weight gain from smoking cessation usually occurs within 5–7 years after smoking cessation [14, 15]. Longer follow-up periods may increase the potential bias for weight gain due to other causes. Thus, we set the follow-up endpoint at the end of 2014 to minimize potential bias in weight gain due to other factors.

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Ningbo University (Approval Number: 2022RS127). The data utilized in this study were obtained from previous medical records. The waiver of informed consent was approved by the Hospital Ethics Committee, as it was determined that this waiver would not adversely affect the rights and health of the subjects. The study was registered with ClinicalTrials.gov (Registration Number: NCT05550688).

Assessment of smoking status and weight change

For each year of follow-up, we identified participants who reported being smokers in the previous year but non-smokers in the current year as recent quitters and considered the current year as the start of smoking cessation.

We calculated the duration of smoking cessation from the start of quitting to the time of relapse, the onset of T2DM, or the end of follow-up. Participants were divided into four groups based on their body weight changes during follow-up: < 0 kg, 0–1.9 kg, 2.0–3.9 kg, and ≥ 4.0 kg. Participants were divided into four groups according to the relative percentage of weight change: < 0%, 0.0–2.9%, 3.0–5.9%, and ≥ 6.0%, respectively, in NAFLD patients. We further assessed the association between the relative percentage of weight change after quitting smoking and the risk of incident T2DM. Additionally, transient quitters were defined as those who reported being non-smokers in the current year but were current smokers in the previous year and next year.

Assessment of demographic and clinical characteristics

Blood samples were measured using an autoanalyzer [21]. Standing height without shoes and body weight in light clothing were measured according to our previous procedure [21]. Blood pressure was determined by an automatic sphygmomanometer while participants were in a resting state. All participants were examined in the morning after fasting overnight.

Diagnosis of NAFLD and T2DM

We diagnosed NAFLD based on abdominal ultrasonography according to the Chinese Liver Disease Association [22]. Abdominal ultrasonography was performed by experienced ultrasonographers [21]. T2DM was defined as individuals with fasting plasma glucose ≥ 126 mg/dl, an HbA1c ≥ 6.5%, or self-reported clinician-diagnosed T2DM [23].

Statistical analysis

Continuous variables are expressed as the mean ± standard deviation and were analyzed by one-way ANOVA or Student’s t test. Categorical variables are expressed as percentages (numbers) and were analyzed using Pearson's chi-squared test.

Cox proportional hazards models were applied to assess the associations of weight change during smoking cessation with the risk of T2DM. The proportional hazards assumption was tested using Schoenfeld residuals. No evidence of violation of the proportional hazards assumption was found. In multivariable models, we constructed three nested models for analysis: (i) Model 1 adjusted for age and body mass index (ii) Model 2, Model 1 plus drinking status and (iii) Model 3, Model 2 plus aspartate aminotransferase, total cholesterol, triacylglycerol, creatinine, albumin, and systolic blood pressure. In the fully adjusted model, we further explored the interaction effect of NAFLD status on the association between smoking cessation and T2DM by fitting an interaction term on a multiplicative scale.

To test the robustness of our findings, we performed three sensitivity analyses: (i) considering the possible synergistic effect of obesity and NAFLD, we further explored the relationship between weight gain after smoking cessation and the risk of T2DM incidence in non-obese individuals at baseline (ii) to reduce the bias of relapse after cessation, we repeated the analysis after removing 376 individuals who relapsed into smoking cigarettes during follow-up and (iii) previous studies and our results evaluated the trajectory of weight change after quitting smoking, with weight gain occurring primarily in the first 2 years after quitting smoking [14]. Thus, we further evaluated the relationship between weight change in the first and second years after smoking cessation and the incidence of T2DM.

Statistical analyses were performed using R version 3.5.1 (The R Foundation for Statistical Computing). P values were two-tailed, and statistical significance was set at P < 0.05.

Results

Clinical characteristics of the study subjects

A total of 6,559 male participants (5150 NAFLD-free individuals and 1,409 NAFLD patients) were included in the cohort analysis at baseline (Supplementary Figure S1). Participants were divided into current smokers and non-smokers according to their baseline smoking status. The clinical characteristics are summarized according to NAFLD status and smoking status in Table 1. Smokers had a greater body mass index, waist circumference, and serum levels of total cholesterol and triglycerides in both NAFLD and NAFLD-free individuals (all P < 0.05), indicating that smoking may be associated with more unfavorable metabolic profiles than non-smokers.

Table 1 Baseline characteristics of the follow-up population with/without NAFLD stratified by smoking status
Full size table

Follow-up outcomes

During a total of 39,259 person-years of follow-up, 222 (15.8%) NAFLD patients and 621 (12.1%) NAFLD-free participants quit smoking. The clinical characteristics are summarized according to smoking status changes during follow-up in Supplementary Table S1. Compared with current smokers, recent quitters were older in both NAFLD and NAFLD-free individuals (both with P < 0.05).

During the follow-up, 363 cases of T2DM were identified (Supplementary Table S2). Compared with participants who did not develop T2DM, those who developed T2DM were older and had greater body mass index and serum triglyceride levels (all with P < 0.05). Notably, smoking was significantly associated with an increased risk of incident T2DM in both NAFLD and NAFLD-free populations (both with P < 0.001 Supplementary Table S3).

Smoking cessation and risk of incident T2DM

The significant association between smoking and an elevated risk of incident T2DM prompted us to investigate whether quitting smoking decreases this risk. Table 2 shows the association between smoking cessation and the risk of incident T2DM among the whole study population. Surprisingly, we found that smoking cessation was significantly associated with an increased risk of incident T2DM compared with current smokers, after adjusting for age, body mass index, drinking status, aspartate aminotransferase, total cholesterol, triacylglycerol, creatinine, albumin, and systolic blood pressure (Fig. 1). The fully adjusted hazard ratio (HR) (95% confidence interval [CI]) for incident T2DM was 2.16 (95% CI 1.67–2.80).

Table 2 Association between smoking cessation and the incidence of T2DM among individuals by smoking status
Full size table
Fig. 1
figure 1

The interaction effect of NAFLD status on the association between smoking cessation and T2DM. CI confidence interval, HR hazard ratio, NAFLD nonalcoholic fatty liver disease, T2DM type 2 diabetes. The Cox proportional hazards model was adjusted for age, body mass index, drinking status, aspartate aminotransferase, total cholesterol, triacylglycerol, creatinine, albumin, and systolic blood pressure. **P < 0.01, ***P < 0.001

Full size image

Because NAFLD is a major risk factor for T2DM, we separately analyzed the association between smoking cessation and the risk of incident T2DM in NAFLD and NAFLD-free participants (Fig. 1). We found that smoking cessation was significantly associated with an increased risk of incident T2DM in NAFLD patients but not in NAFLD-free individuals. The fully adjusted HRs (95% CIs) were 3.23 (2.21–4.74) and 1.30 (0.89–1.90) in NAFLD- and NAFLD-free individuals, respectively. Although the explanation remains unclear, the above findings clearly illustrated that NAFLD patients and NAFLD-free individuals had different risks of incident T2DM after smoking cessation (Pinteraction < 0.001).

Post-cessation weight gain and risk of incident T2DM

Since 58.0% of recent quitters had less than 5 years of follow-up, we assessed the mean cumulative weight gain by smoking cessation duration in the first 4 years. As shown in Supplementary Figure S2, recent quitters experienced a mean weight gain of 0.74 kg in the first year and 1.21 kg in the second year after quitting smoking. Compared with those in current smokers and never-smokers or ex-smokers, weight changes were greatest in recent quitters during the first and second years of follow-up (P < 0.001). These results indicate that individuals who quit smoking showed significantly greater weight gain than did current smokers, never-smokers, or former smokers. The detailed weight changes during the first and second years of follow-up are illustrated in the Sankey diagram (Supplementary Figure S3).

Furthermore, we found that participants with NAFLD experienced greater weight gain after smoking cessation than did those without NAFLD during the follow-up period (2.28 [3.35] kg versus 1.19 [3.37] kg, P < 0.001). As shown in Table 2, compared to current smokers, recent quitters with weight gains of 0–1.9 kg, 2.0–3.9 kg, and ≥ 4.0 kg had a significantly greater risk of incident T2DM among individuals with NAFLD, and the corresponding fully adjusted HRs (95% CIs) were 0.41 (0.06–3.01), 2.39 (1.21–4.70), 4.48 (2.63–7.63), and 6.42 (3.68–11.23), respectively. This indicates that even mild weight gain after smoking cessation significantly increases the risk of T2DM in NAFLD patients.

In addition, we noted a significantly greater risk of T2DM among transient quitters than among current smokers in the fully adjusted model (P < 0.001). This suggests that short periods of smoking cessation may still increase the risk of developing diabetes. Short-term quitters gained more weight after quitting than did continued smokers (0.29 [2.74] versus 0.19 [2.82] in the first year). This further suggested the importance of weight control after smoking cessation.

For NAFLD-free individuals, the fully adjusted HRs (95% CI) for the risk of T2DM were 0.63 (0.27–1.45), 0.87 (0.38–2.01), and 1.54 (0.66–3.56) among recent quitters with weight changes < 0 kg, 0.0–1.9 kg and 2.0–3.9 kg, respectively. We only observed a significant association between ≥ 4.0 kg weight gain after smoking cessation and the risk of T2DM, with a fully adjusted HR (95% CI) of 2.80 (1.54–5.07) (Table 2). This result suggested that mild post-cessation weight gain does not significantly increase the risk of T2DM in NAFLD-free participants.

We further assessed the association between the relative percentage of weight change after quitting smoking and the risk of incident T2DM (Fig. 2). Compared with those of current smokers, the fully adjusted HRs (95% CIs) for incident T2DM were 0.41 (0.06–3.00), 2.23 (1.16–4.28), 5.32 (3.23–8.78), and 6.47 (3.43–12.24) among recent quitters with weight gain < 0%, 0.0–2.9%, 3.0–5.9%, and ≥ 6.0%, respectively, among NAFLD patients. Among NAFLD-free individuals, we only observed a significant association between ≥ 6% weight gain after smoking cessation and the risk of T2DM (P = 0.010). This further confirmed that mild post-cessation weight gain was associated with an increased risk of incident T2DM in NAFLD patients but not in NAFLD-free participants.

Fig. 2
figure 2

Association between the relative percentage of weight change after quitting smoking and the risk of T2DM in the fully adjusted multivariable model. CI confidence interval, HR hazard ratio, NAFLD nonalcoholic fatty liver disease, T2DM type 2 diabetes. The Cox proportional hazards model was adjusted for age, body mass index, drinking status, aspartate aminotransferase, total cholesterol, triacylglycerol, creatinine, albumin, and systolic blood pressure. *P < 0.05, **P < 0.01, ***P < 0.001

Full size image

Sensitivity analyses

To assess the robustness of our findings, we conducted three sensitivity analyses:

First, we explored the association between post-cessation weight gain and the risk of T2DM in non-drinkers and non-obese individuals and observed that the impact of weight gain after smoking cessation was not modified by alcohol consumption or obesity status (Fig. 3A, B).

Fig. 3
figure 3

Sensitivity analyses of the association between smoking cessation and the incidence of T2DM by smoking status in a fully adjusted model. A Among individuals who reported no alcohol consumption at baseline. B Among individuals with a baseline BMI < 28 kg/m2. CI confidence interval, NAFLD nonalcoholic fatty liver disease. The Cox proportional hazards model was adjusted for age, body mass index, drinking status, aspartate aminotransferase, total cholesterol, triacylglycerol, creatinine, albumin, and systolic blood pressure. *P < 0.05, **P < 0.01, ***P < 0.001

Full size image

Second, to minimize the potential impact of smoking relapse, we removed participants who relapsed smoking during follow-up (Supplementary Table S4). Compared with those of current smokers, the fully adjusted HRs (95% CIs) for incident T2DM were 1.00 (0.14–7.30), 3.23 (1.55–6.75), 5.04 (2.69–9.44), and 7.05 (4.04–12.30) for recent quitters with weight gains < 0 kg, 0–1.9 kg, 2.0–3.9 kg, and ≥ 4.0 kg, respectively, among individuals with NAFLD. For NAFLD-free individuals, the corresponding HRs (95% CI) were 0.82 (0.30–2.27), 1.53 (0.61–3.80), 2.50 (0.91–6.88), and 3.60 (1.72–7.55).

Third, we evaluated the relationship between weight change in the first and second years after quitting smoking and the incidence of T2DM (Supplementary Figure S4). We obtained similar results regarding the effect of weight change during the first year after smoking cessation on NAFLD individuals. Compared with those of current smokers, we observed fully adjusted HRs (95% CIs) for incident T2DM of 1.27 (0.57–2.87), 3.98 (2.26–7.02), 4.77 (2.64–8.60), and 7.16 (3.69–13.91) for recent quitters with weight gains < 0 kg, 0–1.9 kg, 2.0–3.9 kg, and ≥ 4.0 kg, respectively, among NAFLD patients. We found a significant relationship between weight gain ≥ 4.0 kg after smoking cessation and T2DM risk (P < 0.001) among NAFLD-free individuals. Similar associations were observed after removing 86 individuals with only 1 year of follow-up (Supplementary Figure S4).

Overall, these sensitivity analyses confirm the robustness of our findings and provide further evidence for the significant association between post-cessation weight gain and T2DM risk in NAFLD patients but not in NAFLD-free individuals.

Discussion

This cohort study provides novel insight into the differential impact of post-cessation weight gain on the risk of incident T2DM in individuals with and without NAFLD. Our findings revealed that even mild weight gain following smoking cessation is significantly associated with an increased risk of T2DM in individuals with NAFLD. In NAFLD-free individuals, this risk becomes apparent only when weight gain exceeds 4.0 kg. These results underscore the unique vulnerability of NAFLD patients to weight gain-related metabolic disturbances after smoking cessation. Furthermore, the sensitivity analyses confirmed the robustness of these findings in subgroups of non-drinkers and non-obese individuals. These observations highlight the critical importance of managing weight gain post-cessation, particularly in NAFLD patients, to mitigate the elevated risk of T2DM and optimize health outcomes.

Smoking is an independent risk factor for NAFLD progression [8, 9]. Jung et al. performed a cohort study of 199,468 Korean adults [9] and reported that current smoking status, pack-years, and urinary cotinine levels were positively associated with the risk of incident NAFLD. Therefore, it is advised that patients with NAFLD quit smoking. Smoking cessation is a common clinical problem. As part of the “Healthy China 2030” initiative, the Chinese government aims to reduce smoking prevalence from 27.7 to 20% by 2030 [24], which means a net reduction of nearly 100 million smokers [25]. However, smoking cessation is often accompanied by weight gain [26], which could be a potential risk factor for NAFLD progression. The magnitude of economic and medical costs associated with incident T2DM in individuals with NAFLD is significant. A better understanding of the impact of smoking cessation on NAFLD may provide new insights into NAFLD pathogenesis and develop preventive strategies.

Previous studies have shown that weight gain after smoking cessation is considered acceptable since it is not associated with an increased risk of cardiovascular disease or chronic diseases [27, 28]. However, our study provides novel and critical insight into the impact of weight gain on the risk of developing T2DM in both NAFLD and NAFLD-free populations. Specifically, we found that even mild weight gain during follow-up after smoking cessation significantly increased the risk of T2DM in individuals with NAFLD. In contrast, for NAFLD-free individuals, an increased risk of T2DM was observed only if they weighed more than 4.0 kg. This critical finding underscores the importance of managing weight gain after smoking cessation, particularly in NAFLD patients, to mitigate the heightened risk of T2DM.

Relapse to smoking after quitting smoking is common among people who have quit smoking [29]. Therefore, we excluded the relapsed smoking population from the sensitivity analysis and again assessed the correlation. In this analysis, we likewise observed a similar phenomenon. These robust results suggest that patients with NAFLD need strict weight control after smoking cessation to reduce the risk of developing T2DM.

Two strengths of this study are the large sample size and its cohort design. Additionally, our study included repeated measures of smoking status during follow-up instead of depending on the “point prevalence” of smoking cessation, which could reflect dynamic changes in smoking status. We further excluded the relapsed smoking population from our sensitivity analysis as mentioned above. The results of these analyses provide evidence for further refining cessation management.

There are some limitations that should be acknowledged. First, our study focused on the Chinese population that regularly underwent health examinations. This may limit the generalizability of our results to other ethnic populations. Second, some additional factors, such as socioeconomic status, physical activity levels, and dietary habits, were not fully adjusted due to the lack of corresponding data. This may have led to residual confounders. Third, female participants were not included in our analysis since female smokers represent a very small minority in Chinese society [30]. The current smoking prevalence was only 2.1% in 2018 [30]. Future studies could include female participants to provide a more comprehensive understanding of the relationship between smoking cessation and type 2 diabetes risk. Fourth, fatty liver was diagnosed by ultrasonographic methods in our study. Although liver biopsy is the gold standard, it is difficult and impractical to perform in community-based studies. The ultrasound definition of steatosis is used in the current guidelines in China [22]. Additionally, since this study was conducted before the introduction of the MASLD definition, we adopted the NAFLD definition for fatty liver. However, according to recent literature, the differences between MASLD and NAFLD are minimal, suggesting that findings based on the NAFLD definition remain applicable and valid under the new MASLD framework [31]. Finally, reliance on self-reported smoking status may introduce bias or misclassification, despite the use of trained nurses to conduct professional and consistent interviews.

In conclusion, this large cohort study indicated that even mild weight gain during follow-up after smoking cessation is significantly associated with an increased risk of T2DM in male individuals with NAFLD. This finding emphasized the importance of weight control after smoking cessation to reduce the potential risk of T2DM, especially in individuals with NAFLD.