Our previous study showed lower all-cause death and cardiac death high BMI group during one year of follow-up10. Analyzing MACCE as the primary outcome and additional meaningful implications were possible using the detailed information included in the KAMIR registry (Table 2A-B). First, group 1 showed a higher incidence of MACCE than group 3 at one year of follow-up, and it was maintained at three years of follow-up. Group 2 also showed poor prognosis in the primary outcome at three years of follow-up, although the difference was not statistically significant at one year of follow-up. The better clinical outcome in the higher BMI groups and the discrepancy of the results according to the follow-up period in group 2 might be attributed to differences in the incidence of CD. (Table 2A-B) The tendency toward increases in the absolute number of MACCE events over time also showed a positive relationship with CD occurrence. High BMI could be considered a protective factor in the occurrence of CD and MACCE because the results remained after adjustment for other confounding factors. Second, higher BMI had a positive effect on the incidence of AD with statistical significance (417 in group 1, 120 in group 2, and 120 in group 3) than that of our previous study (262 in group 1, 261 in group 2, 77 in group 3)10. Third, at one year of follow-up, the incidence of ST was significantly higher in group 1 than in group 3, but not at three years of follow-up. A prior study identified several risk factors related to the incidence of ST17. Statistically significant differences in the baseline characteristics among the groups, such as CKD and heart failure, might have affected ST incidence even after further adjustment. Also, in older people, especially those ≥ 65 years of age, medication compliance could be decreased due to concerns related to adverse reactions with antiplatelet agents18. Compliance with medication-taking might have been poor in group 1 because it was the oldest age (group 1, 69.1 ± 11.7 years; group 2, 63.0 ± 11.6 years; and group 3, 58.1 ± 12.4 years, P < 0.001). Clopidogrel was prescribed more often than ticagrelor or prasugrel in group 1 and vice versa in the case of groups 2 and 3. Different drug potencies could be one reason for the differences during one year of follow-up because taking DAPT for one year after PCI is generally recommended11. The incidence of new-onset HF was higher in group 1 than in group 3, and the proportion of prescription medications was higher in group 3 than in group 1. Therefore, it is necessary to closely monitor cardiac function and prescribe appropriate medications to improve cardiac function and long-term prognosis. The incidence of MI, TVR, and CVA was not significantly different between the BMI groups, perhaps due to the small number of cases. Finally, more minor bleeding events occurred in group 1 than in group 3 at both one and three years of follow-up. The characteristics of the patients in group 1, including low body weight, old age (≥ 65 years), and underlying disease, could predispose them to bleeding events (Table 1).
In subgroup analysis, poor clinical outcomes were identified in the low BMI groups. In particular, low BMI had worse effects on the clinical outcomes in patients without DM than with DM (Fig. 3). It is possible that the accumulation of central fat in DM patients offsets the positive effects of a high BMI19,20. Also, considering that the HbA1C levels were lowest in group 1, there might be few DM patients with low BMIs, and the different number of DM patients between the low and high BMI groups might have influenced the HRs. In addition, the primary outcomes in groups 4 and 5 were not statistically significantly different because the number of patients classified to these groups was insufficient to demonstrate statistically significant results.
The positive effect on clinical outcomes in the high BMI groups could be explained by several theories. First, it is possible that patients classified into the low BMI groups had unhealthy metabolic status with cachexic status. Second, as mentioned above, patients in the high BMI group had a tendency to be actively prescribed medication. The regular prescription of medication and appropriate post-PCI monitoring might have had positive effects on their long-term clinical prognoses.
The study had several limitations. First, it inevitably had the limitations of a nonrandomized retrospective study. Second, it is questionable whether BMI can adequately reflect metabolic status. In previous studies, obesity was divided into metabolically “healthy” and “unhealthy” groups. Total body fat accumulation, especially abdominal fat related to metabolic syndrome, was an important factor in the clinical prognosis21. Because the registry of the study did not include information on peripheral fat deposition, it could act as a confounding factor. However, despite the limitation, the effect of BMI on the primary outcome could have clinical implications, as several other studies reported a positive correlation between BMI and abdominal circumference22,23. An additional limitation is that the proportion of Asian patients with extreme obesity was too small to conclude statistically significant outcomes. Based on the Korea-NIH data, only 0.89% of the general population was classified as having class III obesity24. This is why large-scale studies, including other Asian countries besides Korea, are needed in the future. In spite of these limitations, the study has clinically significant implications as it was a large-scale study in Asians, and it showed three-year long-term clinical outcomes.