Depth of Response was Associated with Progression-Free Survival in Patients with Advanced Non-small Cell Lung Cancer treated with EGFR-TKI

Background: Response Evaluation Criteria in Solid Tumors (RECIST) has been widely utilized to evaluate new therapeutic strategies in cancer. However, RECIST fails to assess the heterogeneity of response in highly active therapies. Depth of response (DepOR), defined as the maximum percentage change in tumor size compared with baseline, may provide a new strategy to evaluate disease response. In the present study, we studied the association between DepOR and progression-free survival (PFS) in patients with advanced non-small cell lung cancer (NSCLC) treated with epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI). Methods: Advanced NSCLC patients harboring EGFR driver mutation (L858R or exon 19 deletion) treated with EGFR-TKI from August 2014 to July 2017 from two sites were retrospetively collected for analysis. Patients were divided into four groups by DepOR (Q1 = 1-25%, Q2 = 26-50%, Q3 = 51-75%, Q4 = 76-100%). Kaplan-Meier curves were plotted for PFS against DepOR and the hazard ratio (HR) was determined through univariable and multivariable cox regression models. Results: In total, 265 patients were included for analysis. The number of patients in Group Q1-Q4 were 91 (34.3%), 73 (27.5%), 65 (24.5%) and 36 (13.6%), respectively. A greater DepOR was significantly associated with a longer PFS (Log-rank P<0.0001). The HRs (95% CI) for PFS comparing patients with different DepOR status were 0.58 (0.42-0.80) for Q2, 0.49 (0.35-0.69) for Q3, and 0.33 (0.22-0.50) for Q4, all compared with patients in Q1. DepOR as a continuous variable was also associated with prolonged PFS (HR, 0.20; 95% CI, 0.13-0.33; P<0.001). Additionally, in the multivariable cox regression model, abnormal LDH, brain metastasis and male were found to be associated with worse PFS outcomes (P<0.05). Conclusion: A greater DepOR is significantly associated with PFS benefit in advanced NSCLC treated with EGFR-TKI, suggesting that it may be a useful clinical outcome to evaluate the response of targeted therapy.


Introduction
Response Evaluation Criteria in Solid Tumors (RECIST) has been widely utilized to evaluate new therapeutic strategies in cancer. Objective response rate (ORR) could be used as an effective surrogate clinical endpoint for overall survival (OS) to predict tumors response via an anatomical approach with a unidimensional measurement of tumor burden according to the RECIST 1 . However, RECIST fails to assess the heterogeneity of response in highly active therapies, or to discern earlier read-out of activity for Ivyspring International Publisher new therapeutics, and it also showed some limitations in predicting OS in some targeted therapies [2][3][4][5] .
In ALEX trial, no significant difference was observed between ORR in non-small cell lung cancer (NSCLC) patients treated with alectinib versus crizotinib (83% vs 76%; p=0.09). However, patients treated with alectinib obtained dramatically prolonged progression-free survival (PFS) compared with those treated with crizotinib 3 . Another trial with afatinib, an epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor (TKI), showed a higher ORR (70% vs 56%; p=0.0083) compared with gefitinib in NSCLC, however, such benefit was not observed for OS 5 . The above results suggested that alternative evaluation tools are needed to supplement RECIST with improved evaluation of clinical benefits in the NSCLC patients treated with targeted therapies. The depth of response (DepOR), defined as the maximum percentage change in tumor size compared with baseline, has been indicated in several clinical trials to fetch up the insufficiency of RECIST in solid tumor evaluations. Increased DepOR were associated with a significantly longer post-progression survival (PPS) or OS in colorectal cancer patients treated with cetuximab or bevacizumab [6][7][8] . Choong-Kun Lee et al. further reported that DepOR could serve as a predictor for long-term outcome in advanced gastric cancer patients treated with trastuzumab 9 . The DepOR has also been investigated in the NSCLC patients treated with ALK inhibitor or anti-PD-1 antibody 10 . Altogether, the studies suggested DepOR to be an effective complement evaluation endpoint for PFS or OS.
Given the rapid development of targeted therapies in advanced NSCLC and the inadequacy of RECIST for monitoring treatment response, in this study, we aimed to provide a new strategy to assess disease response by evaluating the association between DepOR and PFS in NSCLC patients treated with EGFR-TKI.

Data collection
For all patients, the following pretreatment demographic and clinical information were obtained from medical records: age, sex, TNM staging, pathology, EGFR mutation, smoking status, baseline lactic dehydrogenase (LDH), prior treatment, metastatic sites and combined therapy. Target lesion responses were assessed by two independent radiologists via computed tomography (CT) according to RECIST. DepOR was defined as maximum percent change in tumor size compared with baseline, patients were divided into four groups by maximal tumor shrinkage (Q1 = 1-25%, Q2 = 26-50%, Q3 = 51-75%, Q4 = 76-100%). PFS was defined as the interval from the start date of EGFR-TKI administration to the date of progression or death due to any cause.

Statistical analysis
Continuous variables were expressed as means and standard deviations (SDs) if normally distributed or median (range) if not normally distributed. Categorical variables were expressed as number and percentages. Survival probabilities were estimated using the Kaplan-Meier method, and a log-rank test was used to evaluate differences between survival curves. A Cox proportional hazards regression model was used for univariate and multivariate analyses. In the multivariable cox regression model including variables with P value < 0.10 in the univariate cox regression. All P-values were two-sided, and P < 0.05 was considered statistically significant. Statistical analyses were performed with SPSS version 20.

Association between DepOR and PFS
The association between DepOR and PFS was determined among patients in Groups Q1-Q4 that were divided upon the maximal tumor shrinkage as described above. The maximum tumor reduction for each patient was depicted in Figure 1A. The median PFS for patients from Groups Q1-Q4 were 6.8 months, 13.0 months, 13.7 months and 19.4 months, respectively (P<0.05). Kaplan-Meier survival curves were plotted for PFS against DepOR (Figure 1B), showing that a greater DepOR was significantly associated with a longer PFS (Log-rank P < 0.0001).

Univariate and multivariate analysis of progression-free survial of patients treated with EGFR-TKI
In the univariate analysis, DepOR was significantly associated with PFS either as a rank variable (HR, 0.68; 95% CI, 0.60-0.76; P<0.001, Table 2) or as a continous variable (HR, 0.20; 95% CI, 0.13-0.33; P < 0.001, Table 2). In addition, sex, liver metastasis, brain metastasis, surgery prior to the use of TKI, and LDH level were associated or tended to be associated with the progression-free survival. Variables with a P value <0.10 in the univariate anlaysis were subsequently included in the multivariate cox regression model, showing that sex, brain metastasis, LDH level, and DepOR remained to be significantly associated with progression-free survival ( Table 2).   Table 3). After adjusting the potential confounding factors including LDH level (normal vs abnormal), brain metastasis, and sex, DepOR was still an independent prognositc factor for PFS with a HR of 0.57 (0.40-0.80, P<0.001) for Q2, 0.36 (0.25-0.52, P<0.001) for Q3 and 0.32 (0.20-0.51, P<0.001) for Q4 as compared with paitnets in Q1 ( Table 3).

Discussion
In the present study, we have evaluated the relationship between DepOR and PFS in NSCLC patients treated with EGFR-TKI. The results demonstrated that a greater DepOR is associated with PFS benefit from EGFR-TKI treatment. Besides, brain metastasis, abnormal LDH level, and male were associated with worse PFS.
OS is currently the golden standard for oncology clinical trial endpoints 11 , however, it is inefficient and time-consuming for clinical evaluation, thus finding a surrogated early endpoint is important for clinical assessment. ORR evaluated by RECIST criteria has been widely applied by clinicians. However, previous studies revealed that ORR may fail to capture the full benefit of treatment 11 and discern earlier read-out of activity for new therapeutics. In addition, ORR outcome is not always consistent with PFS or OS in evaluating endpoints in oncology 3,4 . Under these circumstances, DepOR has been proposed as a complement factor for ORR to evaluate the clinical benefit 7 .
Investigations have been carried out in several solid tumors using DepOR as a clinical endpoint. In the phase III trial comparing cetuximab in combination with FOLFIRI as first-line treatments in colorectal cancer patients, an increased DepOR, while not ORR, was associated with a longer OS in patients treated with cetuximab plus FOLFIRI 6 . DepOR has also been studied in a post-hoc analysis of CRYSTAL and OPUS trials in colon cancer, in which a greater DepOR was demonstrated to be associated with longer OS and post-progression survival 8 . Another clinical trial showed that gastric cancer patients with a DepOR ≥ 45% obtained longer PFS and OS when treated with trastuzumab 9 . In fact, the satisfactory performance of DepOR as an evaluation factor has not only been shown for targeted therapies, but also for immunotherapy in NSCLC 10 . In addition, DepOR has been investigated in hematologic malignancies to evaluate the changes of M-protein in multiple myeloma and minimal residual disease in leukemia and myeloma [12][13][14][15] , implying the validity of DepOR as a complementary endpoint established for different interventions and different neoplasms.
In the present study, we demonstrated that a greater DepOR was significantly associated with a longer PFS. After adjustment of potential confounding factors including LDH level, brain metastasis, and sex, DepOR was still an independent prognositc factor for PFS. We have proved that DepOR could serve as an effective complementary clinical endpoint to evaluate the clinical benefit of EGFR TKI in patients with NSCLC.
Besides, we showed that an abnormal lactate dehydrogenase level was associated with worse PFS. LDH has been widely used as a classic inflammatory marker for lung cancer patients treated with chemotherapy or targeted therapies [16][17][18] . One previous study demonstrated that an abnormal LDH level might be a negative predictor of survival in NSCLC patients receiving immune checkpoint blockades 19 . However, our study revealed that the abnormal LDH was associated with shorter PFS in advanced NSCLC patients treated with EGFR-TKI, suggesting that LDH may be an independent prognosis indicator for NSCLC instead of a predictive biomarker for immunotherapy.
We also found that brain metastasis was a potential poor prognostic factor independent of the DepOR. Indeed, EGFR-TKIs administrated in this study, including gefitinib, icotinib and erlotinib, have been evaluated in patients with brain metastasis with a mild benefit [20][21][22] , probably because of their poor capability to penetrate the blood-brain barrier. For a majority of the NSCLC patients with central nervous system (CNS) metastasis, brain metastasis was the first site of treatment failure after initial response 23 , supporting the aforementioned blood-brain barrier hypothesis.
Our study has provided a new strategy to evaluate the clinical outcomes, however, there are still some limitations need to be considered. Firstly, this is a retrospective study that may yield selecting bias. Nevertheless, the patient demographics and disease characteristics were well-balanced across the quartiles for most variables. Secondly, the association between DepOR and OS was not analyzed due to the consideration of the changes of treatment for patients upon disease progression, which may introduce potential bias to the results. Thirdly, in our study, the association between DepOR and PFS was conducted only in patients with tumor shrinkage due to the limited number of patients with progressed disease upon EGFR-TKI. Last but not least, the ability of response assessment between DepOR and RECIST was not compared directly, a larger prospective study is warranted for further studies. In summary, we have demonstrated that DepOR was associated with better PFS either as a rank variable or a continuous variable, suggesting that the DepOR may be a supplement to RECIST to evaluate the treatment response in future clinical practice.

Conclusion
The DepOR may be a useful clinical outcome to efficiently evaluate the response of EGFR-TKI in patients with NSCLC. More evidences are needed to understand how to use DepOR as an endpoint in clinical trials.