Racial Disparities in Survival Among Advanced NonSmall Cell Lung Cancer Patients First-Treated with The Combination of Paclitaxel and Platinum

Background: In clinical practice, different chemotherapy schedules for NSCLC 19 have potential impacts on overall survival, but not specific in NCCN guidelines. 20 Although some meta-analyses have been conducted to identify the influencing factors, 21 the results are limited by method. Our study is helpful to supplement the NCCN 22 Guidelines. 23 Methods: In order to evaluate and quantify the potential relationship between 24 demographic characteristics, disease characteristics, etc. and overall survival as well as 25 their impact on objective response rate and safety, we developed parametric 26 proportional hazard regression models to describe the overall survival of advanced 27 NSCLC patients first-treated with the paclitaxel-platinum regimen. 28 Results: The principal finding of this research was that race significantly affected 29 hazard of dying. Hazard of dying was 1.4 times higher in non-East Asians than in East 30 Asians, and the hazard ratio of the OS curve was 0.71(95%CI: 0.65-0.78). The median 31 survival time in East Asians and non-East Asians were estimated as 12.2 (95%CI: 10.5, 32 14.4) and 8.4 (95%CI: 6.5. 11.0) months, respectively. The ORR was 37% (95%CI: 32, 33 41) and 28% (95%CI: 25, 32) in East Asians and non-East Asians, respectively. 34 Conclusions: Paclitaxel-platinum regimen had different efficacy in two racial groups. 35 The developed model suggests that the efficacy and safety of paclitaxel-platinum 36 regimen can vary between different racial populations because of differences in 37 influencing factors (doses, platinum type, cycles, etc.) Oncology Group performance status; MBMA, model-based meta-analysis; NSCLC, 451 non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PFS, 452 progression-free survival. 453 454

with ALK rearrangement, MET mutation, ROS1 rearrangement, and BRAF mutation 48 showing a prevalence of approximately 5%, 4%, 1%, and 2%, respectively. 2 However, 49 more than 50% of patients with advanced NSCLC still have unknown gene mutations, 50 and thus cannot receive targeted therapy. 3 Lee (2014) compared targeted therapy and 51 conventional chemotherapy in EGFR wild-type patients, and the results showed that 52 conventional chemotherapy led to significantly higher progression-free survival (PFS) 53 and objective response rate (ORR) than targeted therapy. 4 In immunotherapy, patients 54 with PD-L1 expression levels above 50% could significantly benefit from conventional 55 chemotherapy; however, in fact, only 23-28% of patients with advanced NSCLC had 56 high PD-L1 expression levels. 5 Therefore, most patients cannot benefit from targeted 57 therapy or immunotherapy, and chemotherapy remains the cornerstone of treatment for 58 advanced NSCLC. 59 The combination of paclitaxel and platinum drugs is a recognized first-line 60 chemotherapy regimen for the treatment of advanced NSCLC. 6 However, the NCCN 61 Clinical Practice Guidelines in Oncology for NSCLC recommend conventional 62 treatment options without clarifying the specific details, such as the times of 63 administration, cycles of chemotherapy, and doses of paclitaxel in combination with 64 platinum drugs, and thus may cause great differences between different paclitaxel-65 platinum regimens in clinical practice. Of particular concern is that the efficacy and 66 safety of a paclitaxel-platinum regimen may not be identical in different populations.   Gao (2012) found that paclitaxel had similar efficacy but better safety when 80 administered once a week versus every 3 weeks, and the weekly schedule was suitable 81 for elderly patients who could not tolerate the standard regimen and for patients with 82 poor performance status scores. 8 In terms of chemotherapy cycles, Soon (2009) found 83 that longer cycles can statistically increase OS and PFS, but did not investigate whether 84 the extension of the cycle would increase the incidence of adverse events. 9 Moreover, 85 this study was published relatively early, and only a few trials were included; thus, there 86 are some potential limitations. Thus far, several studies have indicated that six cycles 87 and three to four cycles of platinum-based first-line treatment did not significantly 88 improve OS and PFS in patients with advanced NSCLC. In addition, the incidence of 89 anemia was higher with six cycles; thus, fewer cycles of chemotherapy were 90 recommended. 10 It should be noted that these studies did not distinguish the 91 heterogeneity of subjects in the trial, and thus their conclusions may be more prone to  In this study, MBMA was conducted to explore and quantitatively assess the Standard error function OS rate is affected by many factors in survival analysis, and censored data are 148 included in many studies. The distribution of OS was not exactly determined nor did it 149 meet the normal distribution and homogeneity of variance. Therefore, hazard rate was 150 used as a dependent variable in this study.

151
In formula (1), h(t) represents the instantaneous hazard of dying at time t, which 152 was obtained by multiplying two parts. Among them, h0(t) describes the base hazard categorical variables were introduced as formula (9): In the covariate screening, the influence of a single covariate on the efficacy end         Table 3. 306 Among the covariates predicted to be related to the regimen, none was found to  In addition, one interesting result emerged that carboplatin tended to be more effective 325 than cisplatin in East Asians, but showed no obvious tendency in non-East Asians.  328 Meta-analysis of ORR between the two racial groups showed that ORR in East 329 Asians was significantly higher than that in non-East Asian patients, which manifested 330 as not overlapping 95%CIs. According to the meta-analysis, the ORR was 37% (95%CI: 331 32, 41) and 28% (95%CI: 25, 32) in East Asians and non-East Asians, respectively.

343
Among non-hematological toxicities, only nervous system disease was significantly 344 different between the two racial groups; its incidence was 3% (95%CI: 1, 5) in East 345 Asians and 9% (95%CI: 6, 13) in non-East Asians. Figure 3C-F presents the meta-346 analysis results of grade 3-4 AEs in the subgroups. There were significant differences 347 in the incidence of AEs between the two racial groups at different covariate levels. In 348 both racial groups, the incidence of leukopenia was significantly higher following six 349 cycles of chemotherapy than after four cycles of chemotherapy. Moreover, the 350 incidence of leukopenia with 28 days per cycle was higher than that with 21 days per 351 cycle in non-East Asians. The incidence of neutropenia following one-time paclitaxel 352 administration per cycle was higher than that after three-time administration.

353
Neutropenia incidence was also different between 28 days per cycle and 21 days per 354 cycle. In East Asians, there were differences in the incidence of leucopenia between 355 cycles of chemotherapy and between platinum types.  In East Asians, carboplatin was more effective than cisplatin, but led to higher 395 incidence of leukopenia. On the contrary, in a study by de Castria (2013) 47 , there was 396 no significant difference in OS between cisplatin and carboplatin. It is possible that this 397 previous finding was due to the lack of distinction between racial groups. In addition, 398 our analysis revealed that four cycles of chemotherapy had better safety and lower 399 incidence of leukopenia than six cycles of chemotherapy.

400
In the meta-analysis of ORR and the incidence of grade 3-4 AEs, meta-regression 401 was used to explore the source of heterogeneity to avoid its influence. Further analysis 402 showed that race, platinum type, and paclitaxel dose were the sources of heterogeneity 403 in ORR, but there was no correlation between these factors. Race was the source of 404 heterogeneity in the incidence of leucopenia and neutropenia. We then performed a 405 subgroup analysis of ORR, leukopenia incidence, and neutropenia incidence in two 406 racial groups. The results showed that the heterogeneity was significantly reduced, but 407 still present. Further meta-regression revealed that platinum type was the source of 408 heterogeneity in ORR among non-East Asian patients, and cycles of chemotherapy 409 were the source of heterogeneity in leucopenia incidence. In addition, platinum type 410 was the source of heterogeneity in leucopenia incidence among East Asians. The indicated that after the data on albumin-bound paclitaxel and liposomal paclitaxel were 427 deleted, the rate of change of model parameter estimates was 4%, which was less than 428 5%. Next, by comparing 95%CI between the simulated efficacy and the original model, 429 we found a high coincidence of these intervals. The most important result was that after 430 the data on Indians, albumin-bound paclitaxel, and liposomal paclitaxel were deleted, 431 race remained as the covariate that significantly affected OS. Our analysis confirms that 432 compared with non-East Asians, East Asians had a better efficacy with paclitaxel-433 platinum regimen. We believe that the findings of this study will help most patients in 434 NSCLC and be helpful to supplement the NCCN Guidelines. At the same time, we 435 suggest that later studies should focus on the differences in efficacy between races. Non-East Asians and East Asians. Therefore, this regimen should be adjusted 443 accordingly when used clinically. East Asians had longer OS and higher 5-year OS rate 444 after treatment with a paclitaxel-platinum-based regimen than non-East Asians. 445 Moreover, the ORR significantly improved. However, the incidence of AEs was 446 relatively high in the East Asian population owing to low tolerance.    Visualization results of ORR and grade 3-4 AEs at different covariate levels in subgroup meta-analysis. Error bars represent 95% CI based on meta-analysis. Cycles, cycles of chemotherapy; Dose, paclitaxel doses, Frequency, times of administration of paclitaxel; One cycle, duration (days) of one cycle;21d, 21days; 28d, 28days; CDDP, cisplatin; CBDCA, carboplatin. Different covariate levels were divided into cycles of chemotherapy (4 versus 6), paclitaxel dose (175 mg/m2versus 200-225 mg/m2,<175 mg/m2 versus 175 mg/m2), times of administration of paclitaxel(1/cycle versus 3/cycle,1/cycle versus 2/cycle), duration (days) of one cycle(21days versus 28days), platinum type(cisplatin versus carboplatin).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.