Incidence, consequences, and predictors of serious chemotherapy‐induced thrombocytopenia in nasopharyngeal carcinoma

Abstract Objectives This study aimed to investigate the incidence, consequences, and predictors of serious chemotherapy‐induced thrombocytopenia (CIT) in nasopharyngeal carcinoma (NPC). Materials and Methods We retrospectively reviewed the clinical records of patients with NPC between 2013 and 2015. Multivariate Cox proportional hazards regression model and propensity score matching were used to estimate the effect of serious CIT on overall survival. Univariate and multivariate logistic regression analyses were applied to identify the predictors of serious CIT. Results and Conclusion The incidence of serious CIT was 5.21% in patients with NPC. Patients who experienced serious thrombocytopenia had a worse long‐term prognosis, while the difference in short‐term survival rate was slight. Chemotherapy regimens of gemcitabine and platinum, 5‐fluorouracil and platinum, taxane and platinum, serum potassium ion concentration, serum lactate dehydrogenase levels, platelet count, red blood cell count, and estimated glomerular filtration rate were predictors of serious CIT.


| INTRODUCTION
In 2020, there were 133,354 new cases of nasopharyngeal carcinoma (NPC), accounting for 0.7% of the total cases of 36 cancers. 1 Radiotherapy and chemotherapy are the main treatment modalities for patients with NPC. A number of clinical trials have indicated that concurrent chemotherapy and induction chemotherapy benefit locally advanced NPC compared to radiotherapy alone. 2 Induction chemotherapy containing platinum, for two to four times, for instance, gemcitabine and cisplatin, docetaxel and cisplatin, cisplatin and 5-fluorouracil, docetaxel plus cisplatin and 5-fluorouracil were recommended. 3 The majority of patients with NPC require chemotherapy because they are likely to be locoregionally advanced when they seek medical care. Several studies have shown that radiotherapy for neck and head cancers does not result in serious thrombocytopenia. [4][5][6] Therefore, it can be deduced that chemotherapy induced serious thrombocytopenia in the treatment of NPC. One study reported that chemotherapy is a risk factor for thrombocytopenia during radiotherapy. 7 Chemotherapy-induced thrombocytopenia (CIT), which has a high cost, leads to reduced, delayed, changed, or discontinued chemotherapy dose. [8][9][10] A platelet count ≤10 × 10 9 /L increases the likelihood of major bleeding. 11 CIT affects not only chemotherapy but also radiotherapy. Therefore, it is necessary to investigate the incidence, consequences, and predictors of serious CIT.

| Patients
We retrospectively reviewed the records of newly diagnosed patients with non-metastatic NPC from the Sun Yat-Sen University Cancer Center (SYSUCC) between 2013 and 2015. The inclusion criteria were as follows: (1) aged from 18 to 75 years when diagnosed; (2) pretreatment white blood cell (WBC) count ≥4 × 10 9 /L and platelet count ≥100 × 10 9 /L; (3) received radiotherapy and platinumcontaining chemotherapy; (4) without other malignant diseases. The exclusion criterion was a change in the chemotherapy regimen for reasons other than thrombocytopenia. According to the Common Terminology Criteria for Adverse Events (CTCAE version 5.0), thrombocytopenia grade 3 is defined as platelet count between 25 × 10 9 /L and 50 × 10 9 /L, and grade 4, platelet count <25 × 10 9 /L. 12 We defined serious CIT as grade 3-4 thrombocytopenia occurring within 120 days from the start of treatment. This retrospective study was approved by the institutional ethics review board of the SYSUCC (approval number B2022-684-01).

| Statistical analyses
To analyze the baseline characteristics of patients, the Mann-Whitney-Wilcoxon test was used for count data and the chi-squared test was used for categorical data. Univariate and multivariate Cox proportional hazards regression models were used to estimate the effect on overall survival (OS), and Schoenfeld residuals were calculated to test the proportional hazards assumption. To further demonstrate the above result, survival analysis was evaluated using the Kaplan-Meier method and examined with the log-rank test after adjusting for confounding factors using propensity score matching (PSM). Univariate and multivariate logistic regression analyses were applied to identify the predictors of serious CIT. To find the threshold of predictors, receiver operating characteristic (ROC) curve analysis was performed, and the area under the curve (AUC) was calculated. OS was defined as the time from diagnosis to death from any cause or the date of last follow-up for patients alive or lost to follow-up. All statistical analyses were conducted using R (version 4.1.3) with tableone, MatchIt, survminer, survival, rms, and pROC packages. [13][14][15][16][17][18] 3 | RESULTS

| Baseline characteristics
A total of 27 patients were excluded from analyses due to a change in chemotherapy regimen for reasons other than thrombocytopenia. The reasons for a change of chemotherapy were as follows: poor response to the first-line chemotherapy (n = 12), serious neutropenia (n = 4), request for a shorter hospital stay (n = 3), and unknown reasons (n = 8). Finally, a total of 3933 patients were included in this study. Table 1 presents the baseline characteristics of patients. The incidence of serious CIT was 5.21% (n = 205) within 120 days from the start of treatment. The incidences of serious CIT among patients receiving platinum alone, gemcitabine and platinum (GP), platinum and 5-fluorouracil (PF), taxane and platinum (TP), taxane plus platinum and 5-fluorouracil (TPF) were 2.66%, 13.97%, 10.17%, 5.02%, and 3.72%.

| Survival analyses after PSM
To further prove the reliability of the conclusion that serious CIT was a risk factor for OS, PSM was applied to balance the other variables, after which the Kaplan-Meier method and log-rank test were used. Variables, such as age, stage, chemotherapy regimen, and LDH level were T A B L E 1 Baseline characteristics before and after PSM. Note: Patients with serious CIT are patients who experienced serious thrombocytopenia within 120 days from the day of diagnosis.

Before PSM
Abbreviations: CIT, chemotherapy-induced thrombocytopenia; GP, gemcitabine and platinum; LDH, lactate dehydrogenase; Platinum alone means that the chemotherapy regimen only includes platinum; PF, 5-fluorouracil and platinum; PSM, propensity score matching; T, tumor; N, node; and stage were regraded based on the eighth edition of AJCC staging system for nasopharyngeal carcinoma; TP, taxane and platinum; TPF, taxane plus platinum and 5-fluorouracil.
included in PSM. Patients with serious CIT were matched to patients without serious CIT for a 3:1 nearest-neighbor PSM with a caliper level of 0.05. A total of 587:201 patients (2.92:1) were included after PSM. The patient characteristics were balanced. Table 1 presents the results. The differences between the variables in the two groups were not significant (p > 0.05).
The median follow-up time, calculated using the reverse Kaplan-Meier method, was 2298 days (95% CI, 2259-2348). The 5-year OS of patients with serious CIT was 82.5% (95% CI, 77.1%-88.2%), which was significantly lower than that of patients without serious CIT, with a 5year OS of 90.9% (95% CI, 88.4%-93.4%). The 1-year OS of patients with serious CIT was 98.98% (95% CI, 97.58%-100.0%), and patients without serious CIT was 99.1% (95% CI, 98.4%-99.9%). Survival curves were described by Kaplan-Meier curves and compared using the log-rank test, as shown in Figure 2. The differences in the survival curves of patients with serious CIT and patients who did not were statistically different (p < 0.05).

| Effects on treatment
To find the underlying cause of decreased OS in patients with CIT, we measured the effects of CIT on chemotherapy and radiotherapy.
Six patients changed the chemotherapy regimen because of CIT and three patients died. 148 patients finished the planned chemotherapy treatment. Fifty-four patients had chemotherapy dose reduction or complete chemotherapy stop. Patients who had chemotherapy reduction or stop had poorer OS (75.9% vs. 83.1%). However, the decrease in OS was statistically insignificant before and after adjusting for important factors (p > 0.05). Sixty-four patients received chemotherapy after CIT and 50% (n = 32) patients had delays in chemotherapy (interval time ≥ 28) due to CIT. The difference in OS was insignificant between patients with and without delays in chemotherapy before and after adjusting for important factors (81.2% vs. 65.6%; p > 0.05). Three patients had unplanned radiotherapy interruption due to side effects other than CIT. Tree patients had incomplete records of radiotherapy duration. One patient who stopped radiotherapy due to CIT received a total dose of 28 Gy for the gross tumor volume and died within 4 years. 172 patients continued radiotherapy despite experiencing CIT. Twenty-nine patients had an interruption of radiotherapy due to CIT. A significantly longer radiotherapy treatment time (RTT) was observed in patients with radiotherapy interruption (44.74 vs. 52.48 days; p < 0.001). However, the effect of radiotherapy interruption on OS was insignificant before and after adjusting for important factors (82.8% vs. 81.4%; p > 0.05). Table 2 presents the results.

| Predictors of CIT
To investigate the predictors of serious CIT, we performed univariate and multivariate logistic regression analyses with 15 variables, including age, sex, stage, chemotherapy regimen, body mass index (BMI), and pretreatment data, including LDH, white blood cell (WBC) count, red blood cell (RBC) count, lymphocyte (LYM) count, platelet (PLT) count, absolute neutrophil count (ANC), serum indirect bilirubin (IBIL), serum potassium ion concentration (K + ), serum sodium ion concentration (Na + ), and estimated glomerular filtration rate (eGFR). The Chronic Kidney F I G U R E 2 Kaplan-Meier curves after PSM. CIT, chemotherapy-induced thrombocytopenia; PSM, propensity score matching. Disease Epidemiology Collaboration creatinine-cystatin C equation was applied to assess eGFR. 20 Table 3 presents the results. Variables with p-value <0.05 in the previous multivariate logistic regression were included in the final multivariate logistic regression. The results of the final multivariate logistic regression, including RBC, PLT, K + , LDH, eGFR, and chemotherapy regimen are shown in Table 3. The variables were lowly collinear, as tested by the variance inflation factor (VIF < 1.8

| DISCUSSION
Serious CIT has long been a hot topic. However, to our knowledge, our study is the first to report the effects of CIT on long-term survival and the first to report that higher serum K + was a risk factor for CIT. Whether patients who experienced neutropenia or leukopenia had worse prognoses was controversial. [21][22][23][24][25][26][27] Prior studies reported a worse prognosis in patients with anemia. 28,29 In our study, grade 3-4 neutropenia, leukopenia, and anemia were not prognostic factors. CIT was a risk factor for OS before and after adjusting the factors of grade 3-4 neutropenia, leukopenia, or anemia which might occur at the same time as CIT. According to Kaplan-Meier curves, the lines were close in the first year and the 1-year OS of patients with serious CIT was 98.98%, and patients without serious CIT was 99.1%. We can infer that CIT will not significantly affect the short-term survival rate. This might be due to the low probability of fatal bleeding because the methods used to treat CIT increase gradually. The 5-year survival rates of the two groups were significantly different.
In our study, the RTT was prolonged in 29 patients with CIT. Previous studies reported a worse OS in patients with prolonged RTT whose threshold was 49-56 days. [30][31][32] The average time of prolonged RTT was 52.48 days in our study. No significant difference in OS was observed in patients with prolonged RTT who had CIT in our study. The negative result was likely due to a brief radiotherapy stop which was shorter than 56 days. According to previous studies, cumulative cisplatin dose >200 mg/m 2 improved survival. 33,34 In our study, 54 patients had chemotherapy dose reduction or complete chemotherapy stop due to CIT. The decrease in OS was insignificant compared with patients who experienced CIT and finished planned chemotherapy. Based on the results, we could conclude that neither the prolonged RTT nor the chemotherapy dose reduction and delays contribute to the decrease in OS. However, it was worth noting that the sample size was small in subgroup analyses. Further studies are needed to investigate the inner relationship between CIT and decreased OS, the effect of thrombopoietic agents on long-term survival, and the microenvironment change when CIT occurs.
In a previous study, lower eGFR (<60 mL/min/1.73 m 2 ) was a risk factor for CIT. 35 In our study, higher eGFR was a protective factor in the univariate and multivariate regression models. The threshold of eGFR was 94.60 mL/ min/1.73 m 2 (AUC, 0.610), which meant the effect of eGFR on CIT also exists in patients with normal renal function. The mechanism might be that better renal function results in more chemotherapy drug excretion, and fewer side effects on megakaryocytes. We also found that eGFR was not a predictive factor of OS through multivariate Cox proportional hazards regression analysis. More chemotherapy drug excretion may not cause a worse prognosis. Further research on the relationship between eGFR and drug concentration in serum is needed to prove this hypothesis.
Our data demonstrated that higher serum K + levels are associated with a higher risk of serious CIT. As already reported, the elevation of serum K + is related to cell and platelet lysis. 36 Previous studies have suggested that potassium supplementation can decrease platelet reactivity. 37 Whether these theories can be applied to serious CIT requires further investigation. A higher level of LDH was a risk factor for CIT in diffuse large B-cell lymphoma. 38 Our study was the first to report that LDH was a risk factor for CIT in solid tumors. The molecular mechanisms for it warrant further research.
Previous studies found that lower RBC count and PLT count were associated with an increased risk of CIT. 39,40 We also proved that higher RBC and PLT were protective factors of CIT. Long-term survival is affected by a serious CIT. This might be an explanation for the conclusion of previous studies that PLT <150 × 10 9 /L was associated with poor survival. 41,42 One study reported some genes associated with the proliferation of megakaryocytes related to CIT in patients with non-small cell lung cancer who have received carboplatin and gemcitabine. 43 Whether the genes above are related to low pretreatment platelet count needs further studies.
Previous studies have reported that PF can cause thrombocytopenia 44 and has more hematologic toxicity. 45 The CIT incidence rate of PF was 4.52%-8% in previous clinical trials. 46,47 In our study, more patients were included, and we found a much higher incidence rate of 10.17%. This was the first time that PF was reported as a risk factor for serious CIT. We found that GP regimen was a risk factor for CIT, which was consistent with the previous study. 48 We found that TP regimen was also a risk factor for CIT whose OR was lower than PF. It is interesting that the three-regimen chemotherapy, TPF, has less toxicity on thrombocytes or megakaryocytes. Further research is required.
It should be noted that we only included patients with NPC from one center, and whether the conclusions could fit other cancers requires further study. Further research is needed to determine whether the results could be applied to patients with pretreatment WBC count <4 × 10 9 /L or PLT count <100 × 10 9 /L.