Three weekly versus weekly concurrent cisplatin: a matter of safety in head and neck cancer


 Background

Radio-chemotherapy with CDDP is the standard for H&N squamous cell cancer. CDDP 100mg/m2/q3 is the standard; alternative schedules are used to reduce toxicity, mostly 40mg/m2/q1.
Methods

Patients were treated from 1/2010 to 1/2017 in two Radiation Oncology Centres. Propensity score analysis (PS) was retrospectively used to compare these two schedules.
Results

Patients analyzed were 166. Most (114/166) had 1w-CDDP while 52 had 3w-CDDP. In the 3w-CDDP group, patients were younger and with better performance status; disease extent was smaller and nodal involvement was more common than in the 1w-CDDP. Acute toxicity was similar in the groups. Treatment compliance was lower in the w-CCDP. OS before PS was better for female, for oropharyngeal disease and for 3w-CDDP group. After PS, survival was not related to the CDDP schedule.
Conclusions

3w-CDDP remains the standard for fit patients, weekly schedule could be safely used in selected patients.


Introduction
From decades, CDDP has been used in the management of locally advanced squamous cell carcinoma of the head and neck (LAHNSCC) in order to enhance the tumoricidal activity of radiation. Among the various CDDP schedules proposed, differing in frequency, dose, and administration, there is level 1 evidence for improvement in loco-regional control and/or overall survival, achieved by three-weekly high-dose (100 mg/m 2 ) cisplatin concurrently with conventional external beam radiotherapy, when compared with radiotherapy alone. The supporting data originate from four large randomized phase III trials investigating the role of cisplatin in both the de nitive and postoperative settings (1,2,3,4) .
Since three-weekly chemotherapy (3w-CDDP) causes signi cant acute toxicity in more than three quarts of patients, many patients are likely to receive sub-optimal cumulative cisplatin dose and dose intensity. This could hamper treatment outcomes and require a proper patient's selection.
Weekly low-dose cisplatin (1w-CDDP) regimes have gained large clinical acceptance, replacing the standard 3w-CDDP schedule at many institutions in daily clinical practice. The background of this choice is the assumption that low-dose, 1w-CDDP could increases treatment compliance maintaining dose intensity and avoiding interruptions of radiotherapy. (5) It could also reduce chemotherapy-related acute and late side effects, facilitate dose adjustments according to clinical conditions during the treatment and therefore outpatient management, with lower hospitalization rates. Several retrospective and small prospective studies (6,7) and different systematic reviews and meta-analysis (8) (9) compared 1w-and 3w-schedules obtaining con icting and inconclusive results, mostly in relation with survival outcomes. Moreover, different prospective randomized trial are actually ongoing in curative setting of both LAHNSCC and nasopharyngeal cancer (NCT03998696, NCT03649048, NCT01171781, JPRN-jRCTs031180135). Weekly 1w-schedules has also be included in de-intensi cation trials for human papilloma virus-related tumors (NCT01530997, NCT01687413). Therefore, waiting for de nitive results, there is an unmet need to provide literature data on homogenous cohorts of patients treated with 1w-CDDP to guide the daily clinical practice.
In this contest, the main objective of this retrospective analysis is to compare, in a real-life setting, two chemotherapy schedules (1w-CDDP 40 mg/m 2 vs 3w-CDDP 100 mg/m 2 ) concomitant to radical radiotherapy in locally advanced head and neck cancers, in terms of acute and overall and relapse free survival. The Propensity Score matched analysis should help to reduce the selection biases that are usually present in a retrospective series.

Materials And Methods
Patients enrolled in this retrospective analysis have been treated between January 1st, 2010 and January All patients had concomitant CDDP-based radical chemo-radiotherapy. Two different CDDP schedules were used in the two Institutions: 100 mg/m 2 every three weeks (3w-CDDP, IEO) and weekly 40 mg/m 2 (1w-CDDP, IRA). In order to reduce the variability related to the patient's body surface differences, the dose was considered as dose/m 2 (ratio of total CDDP dose received by each patient and his/her body surface).
Data were retrospectively collected using a database where all the clinical and therapeutic features were entered.
The ethical committee of the two Institutions approved/noti ed the study.
Stage classi cation was carried out in accordance with the TNM classi cation system, VII Ed. (10) Acute radiation and chemotherapy-related toxicities were analyzed weekly and registered as the higher score occurred during and 3 months after radiotherapy, according to the Common Toxicity Criteria for Adverse Effects (CTCAE) v.4.03.
The RT completion was chosen as reference for survival. Relapse Free Survival (RFS) was the time between the end of treatment and occurrence of local/distant relapse or last follow up, for not relapsed patients; Overall Survival (OS) was the time from the end of radio-chemotherapy to death for any cause or last follow up, for living patients.

STATISTICAL ANALYSIS
The differences between the two treatments were investigated through the χ 2 test.
OS and RFS were calculated through the Kaplan-Meier method and the differences evaluated with the Log-Rank Test.
The Propensity Score matched analysis (PS) (OS and RFS) was introduced to minimize the effect of confounding factors ant to create two homogeneous populations (w-CDDP vs 3w-CDDP). The variables to match the patients (2:1) were age, disease stage and performance status (Karnofsky Performance Status, KPS). At the end, 160 patients were evaluable after the match (114 and 46 patients respectively in the 1w and 3w-CDDp group).
The multivariate analysis was done (OS and RFS), with Cox Regression model, both before and after PS, including all the variables included in the univariate one.
The statistical analysis was made using the IBM® SPSS Statistics® v25.0; the p-values were considered signi cant when p = < 0.05.
Almost all patients had IMRT. (Table 3)

Acute toxicity
The rate of G3-4 acute hematological toxicity was 19.9% in the whole group (18.4% and 23.1% in 1w-CDDP and 3w-CDDP respectively (p = ns). G1-2 anemia and leucopenia were similar in the two groups; G1-2 thrombocytopenia was slightly more frequent in patients treated with w-CDDP (p = 0.01) ( Table 4). Overall G3-4 mucositis, dermatitis and dysphagia rate were 33%, 10.8% and 19% respectively. G 3-4 emesis was higher in the group treated with w-CDDP (p = 0.007) while G1-2 acute xerostomia was more frequent in the group treated with 3w-CDDP (p = 0.009). No severe renal toxicity was recorded (Table 4). Overall Survival Univariate analysis before and after propensity score matched analysis After a median follow-up of 32 months (respectively 35 and 26.5 mm for the 1w and 3w-CDDP), the 1, 2 and 5 yy actuarial OS of the entire series were 97%, 88% and 81.5%. Median OS was not reached neither in entire series nor in the two groups separately (1w and 3w-CDDP).
Before Propensity scored analysis, only female patients showed a statistically signi cant better OS compared with male patients (   (Table 5).
After propensity score matched analysis the statistically signi cant better OS in female patients (p = 0.041) and in oropharyngeal disease (p = 0.047) was con rmed. The worse prognosis of patients with more extensive nodal involvement (N3, p = 0.011) was also demonstrated. Better OS for patients treated with 3w-CDDP was not con rmed (Table 5).
Multivariate analysis before and after propensity score matched analysis The multivariate analysis before the propensity score analysis showed better survival in patients with oropharyngeal cancer, treated with 3w-CDDP and with higher total CDDP/m 2 ( Table 5).
The analysis after PS demonstrated better survival in patients with oropharyngeal disease and low nodal disease burden. None of the therapeutic factors related to chemotherapy or radiotherapy, revealed impact on OS (Table 5).

Relapse Free Survival
Univariate analysis before and after propensity score matched analysis (PS) Mean relapse free survival (RFS) was 69 months (range 63-75 months). Median RFS was not reached neither before nor after the propensity scored analysis.
The results after the propensity score match were almost the same as those registered before applying the propensity analysis. (Table 6) in uences the outcome. The reported better long-term survival, evaluated only on two included papers, could thus be related to this important clinical aspect. About acute toxicity is concerned, the two groups showed the same hematological toxicity (leukopenia, anemia, thrombocytopenia); less frequent severe intestinal toxicity (nausea and vomiting) was registered in the 1w-CDDP group (p = 0.006), whereas severe mucosal toxicity and CDDP delay/interruption were more common in patients with non-nasopharyngeal cancer in the 1-CDDP group (p < 0.0001). As far as treatment compliance is concerned, the data are very heterogeneous, since a signi cant proportion of patients (42% in the weekly CDDP group vs 30% in the three-weekly group) received neo-adjuvant chemotherapy, possibly reducing the tolerance to the concomitant phase. Another limitation of this study is the cumulative analysis of very different disease sites (including nasopharynx) and of different w-CDDP doses (range, 25-40 mg/m 2 /w).
The meta-analysis by Szturz (7) ,including 52 studies, comparing adjuvant/radical 1w-CDDP and 3w-CDDP concomitant to radiotherapy did not show a statistically signi cant difference in OS and relapse rate between the two treatments. Three-weekly administration, however, appeared to be linked with more severe myelosuppression (leukopenia, p = 0.0083 and thrombocytopenia, p = 0.0024), gastrointestinal toxicity (p < 0.001) and severe nephrotoxicity (p = 0.0099), while there were no signi cant differences in mucosal toxicity.
Three-weekly administration was also related to inferior compliance: only 71% of patients completed the full chemotherapy treatment as compared to 88% of the patients who had w-CDDP. It is also worth noting the different distribution of the disease sites in the two groups, with the higher prevalence of oropharynx cancer in the group undergoing three-weekly chemotherapy (49% vs 36%).
A very recent phase III randomized study by Noronha (6) , designed as a non-inferiority study, investigated the outcome of patients with LA head-neck carcinoma (except nasopharynx) treated with 30mg/m 2 w-CDDP compared to the 3w-CDDP 100 mg/m 2 in postoperative/radical setting. The main endpoint of the study was loco-regional control; the secondary ones included toxicity, compliance and OS. The study included 300 patients (150/arm) but 93% were in a postoperative setting (87.3% oral cavity tumors). The 2-year locoregional control was signi cantly higher for the 3w-CDDP (p = 0.014). The results were con rmed after the comparison of patients receiving total CDDP dose > 200 mg/m 2 . As for Progression Free Survival (PFS) and OS, however, no statistically signi cant differences were registered. Regarding toxicity, the 3w regimen was burdened by more frequent severe acute toxicity (p = 0.006) and the hospitalization rate was greater (p < 0.001). The main limitation of this study is the small rate of patients treated radically, due to the preponderance of oral cavity tumors, and the low dose of Cisplatin administered in the weekly schedule (30 mg/m 2 ), compared to the standard of 40 mg/m 2 .
There have also been several attempts to substitute chemotherapy with cetuximab in old and bad general conditions patients, although the Bonner's Study wasn't designed for such patients. (19)(20) The results of these studies are not uniform, but the data of the De-Escalate and RTOG 1016 prospective trials (21)(22) as well as those of a smaller Italian trial (23)(24) with an emphasis on toxicity, did not con rm the hypothesis of the better compliance and equal e cacy of bio-radiotherapy, particularly in patients with better prognosis (HPV positive disease).
In this context, our study aims to contribute to the body of literature on this controversial issue with a retrospective evaluation of the e cacy and tolerability of the two chemotherapy schedules (1w-CDDP 40 mg/m 2 and 3w-CDDP 100mg/m 2 ) administered concurrently with radiotherapy in patients with LA head-neck cancer (oropharynx, hypopharynx and larynx).
The two treatment groups in our series are signi cantly different in relation to patient clinical characteristics (per arm number of patients, gender, age, performance status, alcohol and smoking habits); higher rates of women, young patients and subjects with better KPS and less smoking and alcohol consumption were registered in the 3w-CDDP group. Moreover, in the same group there was a prevalence of oropharynx cancer, even if they had more advanced nodal disease. Nevertheless, the propensity score method applied for the statistical analysis was able to mitigate these inhomogeneities thus rendering more reliable and robust the presented results.
A non-signi cantly higher rate of G3-4 hematologic toxicity was observed for the 3-weekly schedule. No signi cant differences were evident in terms of mucositis or dysphagia. A higher rate of G1-2 thrombocytopenia, mild gastrointestinal toxicity and CDDP interruptions were observed in patients treated with w-CDDP. The higher rate of toxicities could be attributed to the different characteristics of patients treated: more patients with low KPS, older than 70 years and smoke and alcohol addiction were treated with w-CDDP. The subgroup analysis showed that also within subgroup with KPS > = 90 patients of the 1w-CDDP group, are more frequently alcohol and smoking user.
The OS analysis of the present series, not corrected for age, performance status and disease stage, showed a statistically signi cant better survival for patients treated with 3w-CDDP compared to w-CDDP, with 2-and 5years rates of 95.4% vs 84.6% and 95.4% vs. 75.9%, respectively (p = 0.026). This result is, probably, related with a selection bias of the patients in the 3w-CDDP group (younger age, better performance status, less smoking and alcohol consumption and higher rate of HPV positivity). This interpretation of the data is con rmed by the similar survival results registered in the two treatment groups with the propensity scored matched analysis.
The same results have been obtained also for RFS. The results from multivariate analysis after the propensity scored matched analysis, both for OS and RFS showed that neither the interruption of chemotherapy nor the CDDP total dose/m 2 can be identi ed as an independent prognostic factor.
Propensity score analysis is useful to decrease the biases related to the analysis of a non-randomized population, that however cannot be completely eliminated.

Conclusions
Three-weekly CDDP still represents the gold standard in curative and postoperative concurrent chemoradiation for LAHNSCC patients, despite the de nition of the gold standard of the chemotherapy schedule is still much debated.
This is a retrospective, propensity score matched analysis, suggesting the equivalence of the two CDDP schedules in terms of survival outcomes. These data, since they are retrospective in nature, are not per se su cient to modify current clinical practice, but could con rm, together with others already published data, that 1w-CDDP can be safely used in this group of patients. The lower patients' compliance to the 1w-CDDP schedule could be justi ed by the worst patients prognostic factors (older age and lower performance status, alcohol consumption and smoking habits) compared to the 3w-CDDP cohort but it should be taken into account when we choose this personalized approach to support the frailty.

Declarations
Ethics approval and consent to participate: The ethical committee of the two Institutions approved/noti ed the study.
Consent for publication: All patients gave written informed consent for the treatment and anonymous use of their data for educational and research purposes.
Availability of data and material: The datasets analysed during the current study available from the corresponding author on reasonable request.
The other authors have nothing to disclose Funding sources: none Authors' contributions: All submitting authors have had full access to all data and certify to their integrity, and support the decision to submit them for publication.