Lung cancer is the deadliest carcinoma globally, with non-small cell lung cancer (NSCLC) accounting for approximately 80–85% of all lung cancers.1 Overall survival (OS) is considered the most reliable and appropriate endpoint in oncology clinical trials, especially when it can be adequately assessed.2 The OS is accurate and easy to measure due to the easiness of recording the date of death. Additionally, alternative measures, such as tumor shrinkage and progression-free survival (PFS), are considered helpful endpoints in cancer clinical trials because they can be measured earlier and seamlessly and occur more continually than the major endpoint of interest (the ‘true endpoint’).
With the pattern of anticancer therapy in NSCLC shifting to single agents and their combinations, the impact of first-line treatment on OS may be greatly influenced by subsequent therapies.3 In fact, some clinical trial results for NSCLC have reported that prolongation of PFS by first-line chemotherapy does not necessarily affect the prolongation of OS.4 Similar to breast, ovarian, and colorectal cancers 5, 6, 7, the number of drugs available for previously treated patients with advanced NSCLC after first-line chemotherapy is increasing. At the clinical trial level, post-progression survival (PPS) has shown a high correlation with OS following first-, second-, and third-line treatment for metastatic NSCLC.8, 9, 10 In particular, from 2002 to 2012, PPS was reported to be highly correlated with OS, which coincided with the initiation of the use of molecular targeted drugs, such as gefitinib and erlotinib, for metastatic NSCLC.8, 9 A method of assessing PPS, calculating OS as PFS + PPS, was first reported in 2009 by Broglio
The effects of treatments administered after disease progression on survival at the individual level are of great interest. We have previously demonstrated that PPS beyond first- and second-line therapy for NSCLC is strongly associated with OS at the individual level.21 However, the associations of PFS and PPS at the individual level with OS after first-line chemoradiotherapy in patients with locally advanced NSCLC have not been reported to date. Our hypothesis is that the OS of patients with recurrence after chemoradiotherapy may also be strongly related to PPS. Thus, evaluating whether PFS or PPS could have a higher impact on OS beyond first-line chemoradiotherapy in patients with locally advanced NSCLC based on individual-level data may be of practical significance.
Approximately 30% of NSCLC patients have locally advanced lesions that cannot be resected at diagnosis22, and a previous report demonstrated that adding chemotherapy to radiotherapy increased survival benefits.23 A meta-analysis reported that concurrent chemoradiation is the most effective treatment for this patient population24, and, accordingly, chemoradiotherapy is currently recommended as the standard first-line therapy for locally advanced NSCLC.
Stage III NSCLCs are heterogeneous tumours characterized by different levels of nodal involvement. In phase III trials, the median OS of stage III NSCLC patients improved from 12 to 23.3 months.24, 25 Recently, a global phase III trial of dur-valumab
Although numerous studies have been conducted on pre-treated individuals with locally advanced NSCLC, none of the studies related to PPS at an individual level are currently available. Thus, we assessed the correlations of PFS and PPS with OS at the individual level in locally advanced NSCLC cases after first-line concurrent chemoradiotherapy. Moreover, we analysed the prognostic values of various patient characteristics for PPS.
A total of 45 consecutive patients with locally advanced NSCLC who had been treated with first-line concurrent chemoradiotherapy at the Gunma Prefectural Cancer Center between January 2011 and December 2018, and in whom recurrence of the chemoradiotherapy had occurred, were enrolled and retrospectively analysed. Flow chart showing patient selection was shown in Figure 1. The inclusion criteria were as follows: (1) histopathologically or cytologically verified NSCLC; (2) first-line concurrent chemoradiotherapy; (3) treatment with curative intent thoracic radiation > 50 Gy concurrent with platinum-based chemotherapy; and (4) recurrent disease after chemoradiotherapy. The criteria for oligo-recurrence were defined as follows: (1) one or more local/distant recurrences, usually in one or more organs or lymph nodes; (2) disease control at the primary cancer site; (3) one or more distant and local recurrences that can be controlled by local treatment; and (4) no distant or local recurrences other than those controlled in (3).28 The study protocol was approved by the Ethics Committee of the Gunma Prefectural Cancer Center. The protocol was performed in accordance with the 1964 Declaration of Helsinki (revised in 2008). Because of the retrospective nature of the study, the requirement for informed consent from patients was waived, but the opportunity to opt out was guaranteed.
Radiotherapy comprised 6M or 10M X-rays at 2 Gy each, usually five times a week, Monday through Friday. The treatment plan for all patients was based on a three-dimensional treatment planning system; tumour size was determined according to the presence or absence of lymph node metastasis by computed tomography (CT). The clinical target volume was defined and outlined as the tumour volume and lymph node area,
Albumin and serum CRP levels were measured at recurrence after chemoradiotherapy. GPS values were defined as follows: a GPS of 0 (albumin ≥ 3.5 mg/dl and CRP < 1.0 mg/dl), a GPS of 1 (albumin < 3.5 mg/dl or CRP ≥ 1.0 mg/dl), or a GPS of 2 (albumin < 3.5 mg/dl and CRP ≥ 1.0 mg/dl). Tumor response was quantified as the best overall response and maximum tumor shrinkage. Radiographic tumour responses were evaluated using the RECIST version 1.1 as follows: complete response (CR), disappearance of all target lesions; partial response (PR), decrease in the sum of the target lesion diameters by at least 30% compared to baseline diameters; progressive disease (PD), increase of at least 20% in the sum of the target lesion diameters compared to the smallest sum during the study; and stable disease (SD), insufficient shrinkage or expansion to qualify as PR or PD.29 PFS was calculated from the initiation of chemoradiotherapy until PD or death from any cause, and OS was recorded from the first day of chemoradiotherapy until death or was censored on the date of the last follow-up. PPS was recorded as the time from disease progression following the first-line treatment to the date until death or was censored on the date of the last follow-up.
Spearman’s rank correlation and linear regression analyses were performed to determine whether PFS or PPS were correlated with OS. The Kaplan-Meier method was applied to assess survival, and differences were analyzed using the log-rank test. Differences were considered statistically significant at a
The characteristics of the study participants are summarized in Table 1. Of the 45 patients (median age, 71 years; range, 42–82 years) enrolled in the current study, during a median follow-up of 31.5 months (range, 2.6–77.9 months), 29 patients died. CR, PR, SD, and PD were observed in 0, 26, 15, and 4 patients, respectively. The response rate was 57.8% (95% CI: 43.3–72.2), and the disease control rate was 91.1% (95% CI: 82.7–99.4). The median PFS and OS were 10.8 months and 31.6 months, respectively (Figure 2 and Figure 3A, B).
Baseline patient characteristics
Characteristic | N = 45 |
---|---|
Gender | |
Male/female | 33/12 |
Median age at chemoradiotherapy (years) | 71 (41–80) |
Median age at progressive disease (years) | 71 (42–82) |
Performance Status at progressive disease | |
0/1/2/3/4 | 15/22/4/4/0 |
Smoking history | |
Yes/No | 36/9 |
Histology | |
Adenocarcinoma/squamous cell carcinoma/others | 23/16/6 |
Clinical stage at diagnosis | |
IIIA/IIIB/IIIC | 28/14/3 |
Driver mutation/translocation | |
|
6/2/1/0/0/36 |
PD-L1 TPS | |
< 1% / 1–49% / 50%/unknown | 6/5/8/26 |
Progression-free survival (months) | . |
< 6 / 6 | 13/32 |
Overall response to chemoradiotherapy | |
CR/PR/SD/PD/NE | 0/26/15/4/0 |
Glasgow prognostic score (GPS) | |
0–1/2 | 32/13 |
Administration of tyrosine kinase inhibitors | |
Yes/No | 11/34 |
Administration of immune checkpoint inhibitors | |
Yes/No | 12/33 |
Administration of durvalmab | |
Yes/No | 2/43 |
Recurrent pattern | |
Local recurrence/distant metastasis | 17/28 |
Intracranial metastases at recurrence | |
Yes/No | 7/38 |
Liver metastases at recurrence | |
Yes/No | 3/42 |
Bone metastases at recurrence | |
Yes/No | 15/30 |
Oligorecurrrence | |
Yes/No | 11/34 |
Radiotherapy after recurrence (any site) | |
Yes/No | 19/26 |
Number of drug therapies after chemoradiotherapy | |
0/1/2/3/4 | 14/18/9/2/2 |
Median (range) | 1 |
Median (range) radiation dosage (Gy) | 60 (58–70) |
Chemotherapy regimen | |
CDDP + VNR | 1 |
CDDP + S-1 | 0 |
CBDCA + PTX | 30 |
Low dose CBDCA | 14 |
The treatments used after the progression following chemoradiotherapy are shown in Table 2. After chemoradiotherapy, 10 patients did not receive any further treatment, and the median number of subsequent treatments was one (range, 0–4 regimens).
The treatments after post-chemoradiotherapy recurrence
first-line | second-line | third-line | ≧fourth-line | Total | |
---|---|---|---|---|---|
Platinum combination | 11 | 3 | 2 | 0 | 16 |
Platinum combination + ICIs | 0 | 0 | 0 | 0 | 0 |
Docetaxel | 0 | 4 | 2 | 0 | 6 |
Docetaxel+ ramcirumab | 0 | 0 | 0 | 0 | 0 |
Pemetrexed | 0 | 0 | 2 | 0 | 2 |
S1 | 1 | 0 | 0 | 3 | 4 |
Others (single agents) | 0 | 1 | 0 | 0 | 1 |
EGFR-TKIs | 6 | 1 | 0 | 0 | 7 |
ALK-TKIs | 3 | 1 | 0 | 0 | 4 |
ICI monotherapy | 1 | 5 | 2 | 1 | 9 |
Ipilimumab+nivolumab | 1 | 0 | 0 | 0 | 1 |
Investigational agents | 0 | 0 | 0 | 0 | 0 |
Radiotherapy alone | 12 | 1 | 0 | 0 | 13 |
Best supportive care | 10 | - | - | - | 10 |
ALK = anaplastic lymphoma kinase; EGFR = epidermal growth factor receptor; ICI = immune checkpoint inhibitor; S-1 = an oral fluoropyrimidine derivative; TKI = tyrosine kinase inhibitor
The associations between PFS and OS and between PPS and OS are shown in Figure 4A, B. Spearman’s rank correlation coefficient and linear regression analyses showed that PPS was highly correlated with OS (
Since PPS was more strongly associated with OS than did PPS, the next step was to examine the factors influencing PPS. In the univariate analysis (Table 3), histology, driver mutation/translocation, GPS at recurrence (0–1 vs. 2), and liver metastases at recurrence were significantly correlated with PPS (
Univariate Cox regression analysis of patient characteristics for post-progression survival
Median | Post-progression survival |
||||||
---|---|---|---|---|---|---|---|
PPS | Univariate analysis | Multivariate analysis | |||||
(months) | HR | 95% CI | HR | 95% CI | |||
Gender | |||||||
Male/female | 18.1/25.7 | 1.49 | 0.63–4.07 | 0.37 | |||
Age at recurrence (years) | |||||||
< 75 / 75 | 20.0/19.5 | 0.78 | 0.36–1.77 | 0.54 | |||
PS at recurrence | |||||||
0–1 / 2 | 21.3/2.8 | 0.42 | 0.18–1.09 | 0.07 | |||
Smoking history | |||||||
Yes/No | 16.7/25.7 | 1.87 | 0.71–6.39 | 0.21 | |||
Histology | |||||||
Adenocarcinoma/non-adenocarcinoma | 25.7/10.5 | 0.37 | 0.17–0.79 | 1.06 | 0.36–3.04 | 0.90 | |
Driver mutation/translocation | |||||||
Yes/No | 27.3/15.1 | 0.32 | 0.07–0.95 | 0.61 | 0.13–2.23 | 0.47 | |
Best overall response of chemoradiotherapy | |||||||
PR/non-PR | 15.1/22.1 | 1.82 | 0.86–4.11 | 0.11 | |||
Progression-free survival | |||||||
< 6 months / 6 months | 6.4 / 24.4 | 1.97 | 0.89–4.19 | 0.09 | |||
Glasgow prognostic score (GPS) at recurrence | |||||||
0–1/2 | 25.7/6.7 | 0.23 | 0.11–0.52 | 0.2 | 0.06–0.55 | ||
Recurrence pattern | |||||||
Local recurrence/distant metastasis | 40.7/16.7 | 0.45 | 0.18–1.03 | 0.05 | |||
Intracranial metastases at recurrence | |||||||
Yes/No | 6.7/21.3 | 1.75 | 0.64–4.09 | 0.25 | |||
Liver metastases at recurrence | |||||||
Yes/No | 4.2/21.3 | 6.82 | 1.50–22.8 | 12.7 | 2.40–56.8 | ||
Bone metastases at recurrence | |||||||
Yes/No | 18.1/20.0 | 1.40 | 0.60–3.06 | 0.41 | |||
Oligorecurrence at recurrence | |||||||
Yes/No | 22.1/19.2 | 0.77 | 0.30–1.75 | 0.55 |
Values in bold are statistically significant (
Next, log-rank tests demonstrated that PPS has a different prognosis for patients according to GPS at relapse (0–1 vs. 2) (
Here, we assessed the association between OS and PFS and between OS and PPS after first-line chemoradiotherapy at the individual level and elucidated that PPS was highly correlated with OS, whereas PFS was weakly correlated with OS. Furthermore, GPS and liver metastases at recurrence were found to be independent prognostic clinical factors for PPS.
The usefulness of alternative endpoints has been demonstrated by several meta-analyses30,31, and biostatisticians have previously reported a variety of alternative endpoints.32,33 In extensive-disease small cell lung cancer, response to treatment and PFS have been proposed as valid alternative endpoints to OS34, but their potency is disputable in advanced NSCLC.35 Broglio
In this population of patients treated with chemoradiotherapy, PPS had a strong effect on OS, but PFS did not have a sufficient effect on OS. Moreover, we demonstrated that PFS was shorter than PPS; thus, PPS was more strongly correlated with OS than PFS, with a linear PPS-OS correlation (Figure 4A, B), which is evidenced by the large
Based on trial-level data for the first-line treatment of advanced NSCLC, a favourable PS and administration of first-line monotherapy and molecular targeted therapy are associated with a longer PPS.8 In addition, individual-level data of patients with postoperative recurrence of NSCLC show that PPS is influenced by PS at recurrence and the use of tyrosine kinase inhibitors (TKIs).38 Several reports have also demonstrated that PPS is highly associated with OS after first-line chemotherapy and that factors affecting PPS include PS and response to chemotherapy.39, 40, 41 However, the factors influencing PPS based on individual-level data after first-line chemoradiotherapy for patients with locally advanced NSCLC are not well understood; thus, we have further attempted to explore the clinical factors influencing PPS.
We found that the GPS (0–1/2) and liver metastases at recurrence (presence/absence) were highly associated with PPS, and we confirmed these associations using log-rank tests. The patient cohort with a GPS of 0–1 had a significantly longer PPS than that with a GPS of 2. In addition to disease stage and conventional prognostic factors, GPS has been demonstrated to be useful in determining the prognosis of lung cancer.14, 15, 16, 17, 18, 19 The GPS is composed of albumin and serum CRP levels, and these clinical parameters are monitored in clinical practice. In the present study, multivariate analysis demonstrated that GPS, but not PS, correlated independently with PPS. PS is a subjective scoring system that evaluates the overall general condition of cancer patients. In the present study, univariate analysis showed a trend towards better PPS for progressive disease in patients with PS 0–1 than in those with PS 2 or higher, but there was no significant difference. PS has been demonstrated to be a potent prognostic factor11,12, but even for patients treated with chemoradiotherapy in the present cohort, PS at recurrence did not have a significant impact on the disease course. In contrast, GPS is an objective and reproducible parameter, which is useful for a more accurate classification of patients by the three-index rating system. Consequently, GPS may be suitable for clinical pre-treatment evaluations. Furthermore, in the current study, the presence of liver metastasis at the time of recurrence was an independent prognostic factor for PPS. Previous studies reported that NSCLC patients with liver metastases have a poor prognosis.42, 43, 44 However, the number of cases with liver metastasis in the present study was small, and our findings need to be confirmed using a larger sample. Moreover, the presence of driver gene mutation/translocation was statistically significant for PPS in the univariate analysis but not in the multivariate analysis. The reason for this outcome is unclear but may be due to the small size of the patient population. In order to resolve the reason for this, we believe that it is necessary to conduct a study with a larger sample size.
Notably, the number of chemotherapeutic regimens for disease progression after first-line chemoradiotherapy is increasing primarily due to the development of more anticancer agents, such as docetaxel, pemetrexed, oral fluoropyrimidine derivative S-1, TKIs, and immune checkpoint inhibitors (ICIs), available for further-line treatment of metastatic NSCLC. As shown in Table 2, various anticancer agents were administered to the patient population in the current analysis. Durvalumab was used in two patients as maintenance therapy after chemoradiotherapy. Maintenance therapy with durvalumab has been reported to improve the prognosis after concurrent chemoradiotherapy26,27 and is currently the standard of care; it has been used in clinical practice in Japan since 2018. The patients included in the current study were from an earlier era when durvalumab was not the standard care. Our findings may lead to high expectations for PPS after durvalumab use in clinical practice, and it was meaningful to include many cases before the use of durvalumab in the present study. Cytotoxic anticancer drugs have been reported to be highly effective after ICI use. For example, docetaxel plus ramucirumab demonstrated a higher response rate when administered after ICI failure compared to treatment regimens without prior ICI use.27 The aforementioned treatment sequence may vary according to the clinical practice guidelines for durvalumab. In the future, it will be important to conduct a similar study on patients who have received durvalumab to determine if our findings will be replicated.
This study has several limitations. The number of patients included in the analysis was relatively small. However, since the number of patients with locally advanced NSCLC treated with first-line concurrent chemoradiotherapy is limited at any facility, the problem of this limitation is difficult to resolve as the aim of this analysis was to evaluate cases with a similar treatment background. Notwithstanding, our facility treats a relatively large number of patients with locally advanced NSCLC, and we have a fairly consistent treatment strategy and follow the standard care guidelines. Despite the possibility of bias due to the single-center nature of the study, understanding the nature of this bias can allow us to make a clinical sense of the results. Second, the point of disease progression might have varied because each physician decided when to record the response and disease progression. However, this variability is considered a limitation of all retrospective studies and is difficult to resolve and should be taken into account when interpreting the results.
In conclusion, our analysis of individual-level data for first-line chemoradiotherapy demonstrated that PPS was highly correlated with OS in patients with locally advanced NSCLC. Furthermore, GPS and liver metastases at recurrence were found to be independent prognostic factors for PPS. Thus, we conclude that the treatment course for disease progression after first-line chemoradiotherapy has a significant impact on OS, and the clinical significance of these findings should be verified in a larger patient cohort for generalizability to other patient populations.