Prognostic Evaluation of the Charlson Comorbidity Index for Breast Cancer Patients by Propensity Score Matching Analysis.

INTRODUCTION
With the aging of the society, the number of elderly patients with cancer is increasing. As the population continues to age, the prevalence of comorbidities observed in oncology practice is also expected to increase. Therefore, understanding the impact of comorbidities on cancer care and management strategies is essential [1].
Breast cancer is the most common cancer among women worldwide [2], and the number of patients with breast cancer with comorbidities continues to increase. In patients with early-stage breast cancer undergoing combined multimodal treatment, comorbidities significantly affect both treatment strategies and prognosis.
The Charlson Comorbidity Index (CCI) was first proposed by Charlson et al. [3] in 1987. Several studies have reported the predictive value for assessing the relationship between comorbidities and survival in various cancers [456789]. Additionally, numerous studies have evaluated how comorbidities affect breast cancer prognosis [10]. Previous studies using the CCI have demonstrated that a CCI score of 1 is associated with a significantly higher risk of death compared with a CCI score of 0 [111213]. However, comorbidities often increase with age, and prognosis inevitably worsens with advancing age. Therefore, increased comorbidities and age are strong confounders of the prognosis of breast cancer. To eliminate this effect, some studies have focused on evaluating the impact of CCI scores on breast cancer prognosis in elderly patients [1415]. To eliminate the influence of age and treatment intensity, we analyzed comorbidities and breast cancer prognosis using propensity score matching (PSM).

METHODS

Patients
A total of 1,403 patients who were diagnosed with primary breast cancer and underwent curative surgery at Kagoshima University Hospital (Kagoshima, Japan) between September 1992 and December 2023 were enrolled. Clinicopathological factors were compared between patients based on the CCI scores. Neoadjuvant chemotherapy and adjuvant therapy regimens were selected for each patient in accordance with the National Comprehensive Cancer Network guidelines and the Japanese Breast Cancer Society Clinical Practice Guidelines [1617]. The seventh edition of the TNM classification was used to determine staging [18]. This study was approved by the Institutional Review Board of Kagoshima University Hospital (#240083-Eki). Informed consent was obtained through an opt-out process on the hospital website.

Data acquisition and CCI score
The comorbidity status of each patient was recorded in the medical records before surgery for breast cancer, based on self-reported medical history and previous medical records. Comorbidity status was evaluated according to the CCI score criteria (Supplementary Table 1) [3]. The CCI considers solid tumor incidence; however, breast cancer itself was excluded from the calculation. Therefore, the CCI used in this study represents a modification tailored for our analysis. Although this modification is not standard, it has been applied in a previous study [4]. Metachronous breast cancers and solid cancers other than breast cancer were included in the CCI calculation if they were active or diagnosed within 5 years of breast cancer surgery.
Estrogen receptor (ER) and progesterone receptor (PgR) positivity was evaluated by immunostaining breast tumor cell nuclei in both noninvasive and invasive tumor samples. A rate of ≥ 1% positive cells was considered positive. Human epidermal growth factor receptor 2 (HER2) status was evaluated only in samples with tumor cell infiltration and was defined as positive when the immunohistochemistry (IHC) score was 3+ or when an IHC score of 2+ was accompanied by a positive finding by fluorescence in situ hybridization. Ki-67 status was evaluated only in samples with tumor cell infiltration, and the cutoff value was set at 20%.
Breast cancer symptoms were extracted from the medical records as the presence or absence of symptoms reported by the patients. Most symptoms involved lump awareness but also included nipple discharge, skin dimpling, pain, and other breast symptoms. Cases without symptoms are detected through imaging tests, such as mammography (MMG) screening or computed tomography scans. “MMG screening” in the table refers to breast cancer diagnosis resulting from MMG examinations conducted as part of public or corporate cancer screening programs.
Recurrence included both local and distant recurrences but excluded metachronous breast cancer.

Statistical analysis
One-to-one PSM was performed using the nearest-neighbor method with a 0.2 caliper. Covariates for PSM included age at surgery, year of diagnosis, and initial T and N stages. After matching, the disease-free survival (DFS), breast cancer-specific survival, and overall survival (OS) data were compared using the CCI score. The log-rank test was used for the significance test of the Kaplan-Meier curves. All p-values were two-sided, and a p-value of < 0.05 was considered statistically significant. All statistical analyses were conducted using JMP Pro, version 16.1.0 for Mac OS (SAS Institute Japan Ltd., Tokyo, Japan).

RESULTS

Patient characteristics
A total of 1,403 patients who underwent curative surgery for breast cancer were included in this study. Among them, 392 patients had a CCI score of ≥ 1. Compared with patients with a CCI score of 0, those with a score ≥ 1 were significantly older, had a significantly better N factor (absent versus present), and had a significantly lower TNM stage (0–1 vs. 2–3). They also exhibited significantly lower Ki-67 values (≤ 20 vs. >20) and were significantly less likely to experience breast cancer recurrence. Furthermore, patients with a CCI score ≥ 1 had significantly fewer breast cancer symptoms and were significantly less likely to have tumors identified on screening MMG (Table 1).
Due to marked selection bias in our cohort, a PSM analysis was performed. After PSM, 764 patients were included (Figure 1). Most covariates associated with breast cancer prognosis, including T factor, N factor, TNM stage, ER status, PgR status, and HER2 status, were balanced between the groups (Table 2). The only frequency of screening MMG for identifying breast cancer was significantly lower in patients with a CCI score of ≥ 1 after PSM. No differences in perioperative treatment were observed between patients with CCI scores of 0 versus ≥ 1 after PSM (Table 3).

Survival outcomes
After PSM, to identify the cutoff CCI score that informs the prognosis of breast cancer, the prognosis was compared between the two groups according to several CCI cutoff scores. DFS, OS, and breast cancer-specific OS were analyzed for each grouping using CCI cut-off values of 1, 2, and 3 for each survival parameter.
No significant differences in DFS or breast cancer-specific OS according to the CCI score were observed (Figure 2A-F). Additionally, no significant difference in OS was observed between patients with a CCI score of 0 versus ≥ 1 or between those with a CCI score of 0 or 1 versus ≥ 2. However, OS was significantly worse in patients with CCI scores ≥ 3 compared with those with scores of 0–2 (Figure 2G-I).
Comorbidities and causes of death in patients with CCI scores ≥ 3 are summarized in Table 4. Covariates correlated with breast cancer prognosis, including age, year of diagnosis, T factor, N factor, TNM stage, ER status, PgR status, and HER2 status, were not significantly different between CCI 0–2 and CCI ≥ 3 groups (Table 5).

DISCUSSION
In this study, we evaluated the impact of comorbidities on the prognosis of patients with breast cancer. The CCI was used to assess comorbidities. The CCI has undergone modifications, such as the age-adjusted CCI (ACCI) proposed by Charlson et al. [19] in 1994. An advantage of the ACCI is that it accounts for the shorter life expectancy of elderly people; however, it cannot distinguish between the influence of age and comorbidities. In general, the older a person, the more comorbidities they are likely to have. Herein, we used the CCI to assess the direct effect of comorbidities on the prognosis of patients with breast cancer.
In our cohort, patients with a CCI score of ≥ 1 had tumors with a significantly lower T factor, a lower N factor, an earlier stage, and an association with a lower recurrence rate compared with patients with a CCI score of 0 (Table 1). Moreover, patients with a CCI score of ≥ 1 had significantly less frequent MMG screening compared with patients with a CCI score of 0. Because our cohort contained a significant amount of bias, to eliminate selection bias, we selected patients using PSM. Although this study was a retrospective analysis and included some bias, no significant differences were observed in factors affecting prognosis, such as age, year of diagnosis, stage, subtype, and breast cancer treatment after PSM (Tables 2 and 3). Therefore, it is reasonable that no significant differences in DFS or breast cancer-specific OS were observed according to CCI score. Regarding OS, no significant differences in OS according to CCI scores of 1 and 2 were observed, while significant differences in OS between patients with a CCI score of ≥ 3 and other group comparisons were observed. This finding indicates that patients with CCI scores ≥ 3 had a significantly higher rate of deaths due to diseases other than breast cancer. Notably, no significant differences were observed in clinicopathologic factors affecting breast cancer prognosis, including age, year of diagnosis, T factor, N factor, stage, ER status, and HER2 status and components of adjuvant therapy between patients with CCI scores of 0 to 2 and those with CCI scores ≥ 3 after PSM (Table 5). Previous studies have shown that the hazard ratios for both breast cancer-specific mortality and all-cause mortality increase with higher CCI scores. In a previous study that analyzed breast cancer outcomes in the elderly using PSM, a significant decrease in survival rate was observed among those with an ACCI score ≥ 3 [7]. However, the study used the ACCI and included the natural consequence that older people have shorter survival times; it remains unclear how the severity of comorbidities alone affects breast cancer mortality. Furthermore, the effects of the CCI score on prognosis in patients receiving the same treatment regimen remain uncertain, as older people often avoid intensive treatments such as systemic chemotherapy [20]. In this study, PSM allowed the evaluation of the prognostic impact of the CCI score in patients with breast cancer while minimizing the effects of age and intensity of breast cancer treatment. This finding suggests that when treatment conditions are comparable, the risk of breast cancer recurrence or mortality does not differ between patients with low and high CCI scores. However, for patients with breast cancer with CCI scores ≥ 3, treatment strategies and follow-up schedules should be carefully designed to balance treatment intensity with effects on comorbidities, because these comorbidities are likely to determine the overall prognosis determinant. A previous study indicated that even when receiving the recommended treatment, patients with an extremely high mortality risk due to comorbidities may not benefit from breast cancer therapy [9]. Our results suggest that patients with CCI scores ≥ 3 represent a patient group that fits this description.
The limitations of this study include its single-center, retrospective design, relatively small size, and extended observation period.
In conclusion, deaths in patients with breast cancer with CCI scores ≥ 3 are often due to diseases other than breast cancer.