Impact of prior cancer history on the survival of patients with larynx cancer

Data was obtained from the custom SEER database [Incidence- SEER 18 Regs Custom Data (with additional treatment fields), Nov 2018 Sub (1975–2016 varying)], which collected cancer data that covers approximately 28% of the United States population [18]. The SEER*Stat software version 8.3.6.1 (National Cancer Institute, USA) was utilized to access the data from SEER database. The patients who were diagnosed with larynx cancer (site code C32.0, C32.1, C32.2 C32.3, and C32.8) from 2004 to 2015 were extracted from the database. Only patients with only one primary tumor or patients who had exactly one prior malignancy were included. In order to exclude the synchronous primary cancers, a 2-month time interval was required between the prior cancer diagnosis and the second larynx cancer diagnosis. Other exclusion criteria were listed as follows: 1) patients with incomplete follow-up; 2) patients whose diagnosis was made at autopsy or based on a death certificate; 3) and patients who had unknown information of diagnosed months. Finally, a total of 24,812 eligible patients were included in this study. The detail flowchart of case selection can be seen in Fig. S1.

The analysis involved multiple variables including demographic characteristics (year of diagnosis, age at diagnosis, gender, race, and marital status), disease characteristics (histologic grade, AJCC stage, and prior cancer type), and treatment characteristics (surgery, radiotherapy and chemotherapy). Specially, the continuous variables, age at diagnosis, were transformed into categorical variables (< 45, 45–64, and > 64). Marital status including divorced, single, widowed, separated and domestic partner were classified into other status. The history of prior cancer was determined from the SEER sequence numbers code, which records the sequence of all malignancies occurred over the lifetime of the patient. Based on the recode of year of diagnosis and month of diagnosis, we calculated the time interval between two cancer record by subtracting the diagnosis date of the prior cancer from that of index larynx cancer. Vital status record were utilized to calculate the overall survival (OS). In addition, patients who died directly from larynx cancer were recorded as dead (attributable to this cancer) based on the code of cause-specific death classification in the SEER program. Alive or dead from a cause other than larynx cancer were considered as a competing risk event and censored at the time of the event. Hence, the SEER cause-specific death classification was utilized to calculate the larynx cancer-specific mortality (LCSM).

Based on the SEER sequence number recode, we categorized the patients into two groups, including patients with primary larynx cancer (PLC) and patients having larynx cancer as the second primary malignancy (SLC). Descriptive statistic was utilized to summarize the demographic and clinical factors of these two groups of patients. Clinicopathologic characteristics between patients with or without prior cancer were compared using Pearson chi-square tests. The Log-rank test or Gray’s test was used to compare differences of OS and LCSM between patients with or without prior cancer record before and after propensity score matching (PSM). The detailed description for PSM could be seen in Supplementary material. The multivariate Cox regression analysis was used to identify whether the prior cancer history was an independent prognostic factor for OS. Meanwhile, multivariable Fine and Gray competing-risk regression model was built to analyze the impact of prior cancer history on LCSM. The common demographic and clinical characteristics, including age, gender, race, marital status, AJCC stage (sixth edition), grade, and treatment modalities (surgery, chemotherapy, and radiotherapy) were included as covariates.

Descriptive statistic, Pearson Chi-square test, and Cox proportional hazards model were performed using SPSS 24.0 (IBM Corp). Kaplan-Meier plot, log-rank test, and the Fine and Gray competing-risk regression model were plotted or conducted by using R software version 4.0.0.

A total of 24,812 eligible patients with larynx cancer were included, of whom 2436 (9.8%) had a history of prior cancer. 67.3% of the patients had their prior cancer identified as AJCC stage I-II. The median time interval (IQR) between the index larynx cancer and the prior cancer was 63 (29–112) months, and the median follow-up (IQR) was 26 (11–58), starting from the diagnosis of larynx cancer (Table 1). After adjusted for propensity scores, all variables were well-balanced between patients with or without prior cancer (Table 2).

Prostate (36%), lung and bronchus (10%), urinary bladder (7%), and breast (6%) were the most common prior cancer types (Table 3). The overall survival (OS) varied significantly among larynx cancer patients who had different types of prior cancer (P < 0.001). Patients with prior lung cancer had the shortest OS, whereas those with prior prostate or colon cancer had a relatively good prognosis (Fig. 1).

As shown in Fig. 2a and Fig. S3A, patients with prior cancer history had significant inferior OS but decreased LCSM compared to those patients without prior cancer. After adjusted for propensity scores, a similar result was shown in Fig. 2b and Fig. S3B. Then, subgroup analyses were conducted for patients stratified by timing and types of prior cancer. Figure 3 indicated that patients with a prior cancer diagnosis had significant adverse impact on OS regardless of the time interval between the index larynx cancer and the prior cancer diagnosis. For LCSM, our results showed that patients with a prior cancer history had a marginal or significantly decreased LCSM compared to patients without a history of prior cancer, especially when the time interval was larger than 12 months (Fig. S4). As shown in Fig. 4 and Fig. S5, the impact of prior cancer on the OS and LCSM of larynx cancer patients varied extensively with different cancer types. The result showed that a prior prostate cancer diagnosis had no significant impact on the OS of larynx cancer patients, but exerted a significant protective factor for LCSM (Fig. 4a and Fig. S5A). On the contrary, larynx cancer patients with a prior lung or breast cancer diagnosis had dramatically inferior OS as compared to patients without a prior history of cancer, but not the LCSM (Fig. 4b and d, and Fig. S5B and D). Moreover, larynx cancer patients with a prior history of urinary bladder cancer or colon cancer both had similar OS and LCSM in comparison with patients without a prior cancer history (Fig. 4c and f, and Fig. S5C and F). Besides, a prior larynx cancer showed an inferior OS as well as increased LCSS, as compared to patients without a prior cancer diagnosis (Fig. 4e and Fig. S5E).

As shown in Table 4, the multivariate Cox analysis for patients’ OS indicated that a history of prior cancer was an independent risk factor for OS (AHR =1.30; 95% CI [1.21–1.41]). On the contrary, the multivariate competitive risk analysis showed that a history of prior cancer was an independent protective factor for LCSM (AHR = 0.83; 95% CI [0.72–0.94]). This data, combined with the OS result from the Cox analysis, suggested that patients with prior history of cancer were more likely to die from non-larynx cause than those without prior cancer. In addition, the Cox model suggested that a married status and treatment modalities including surgery, and radiotherapy were significant protective factors for OS of larynx patients with prior cancer history, whereas later AJCC stage, and high histological grade were risk factors. The competing risk model determined that a married status and treatment modalities like surgery and radiotherapy were protective factors for LCSM of patients with prior cancer history (Table 4).

Subgroup analyses were further performed to compare the OS and LCSM of larynx cancer patients with or without prior cancer history. As shown in Fig. 5a, a prior cancer history was significantly associated with shorter OS in most subgroups, except in patients with younger age (< 45), non-white race or higher histological grade (III-IV). However, regarding to LCSM, a prior cancer history tended to act as a protective factor in most subgroups (Fig. 5b).