Real-world treatment patterns for patients receiving second-line and third-line treatment for advanced non-small cell lung cancer: A systematic review of recently published studies

Most patients with advanced non-small cell lung cancer (NSCLC) have a poor prognosis and receive limited benefit from conventional treatments, especially in later lines of therapy. In recent years, several novel therapies have been approved for second- and third-line treatment of advanced NSCLC. In light of these approvals, it is valuable to understand the uptake of these new treatments in routine clinical practice and their impact on patient care. A systematic literature search was conducted in multiple scientific databases to identify observational cohort studies published between January 2010 and March 2017 that described second- or third-line treatment patterns and clinical outcomes in patients with advanced NSCLC. A qualitative data synthesis was performed because a meta-analysis was not possible due to the heterogeneity of the study populations. A total of 12 different study cohorts in 15 articles were identified. In these cohorts, single-agent chemotherapy was the most commonly administered treatment in both the second- and third-line settings. In the 5 studies that described survival from the time of second-line treatment initiation, median overall survival ranged from 4.6 months (95% CI, 3.8–5.7) to 12.8 months (95% CI, 10.7–14.5). There was limited information on the use of biomarker-directed therapy in these patient populations. This systematic literature review offers insights into the adoption of novel therapies into routine clinical practice for second- and third-line treatment of patients with advanced NSCLC. This information provides a valuable real-world context for the impact of recently approved treatments for advanced NSCLC.

Introduction Brief background on NSCLC Globally, lung cancer is the second most common newly diagnosed cancer and the leading cause of cancer death. In 2012, there were an estimated 1.8 million new lung cancer cases and almost 1.6 million deaths [1]. Non-small cell lung cancer (NSCLC) accounts for approximately 83% of all newly diagnosed lung cancers, and most patients (70%) are diagnosed with advanced disease [2]. Despite treatment advancements, overall survival remains poor with 5-year survival estimates globally ranging from 10% to 20% [3][4][5].
Lung cancer imposes a substantial economic and societal burden [6]. In Europe, lung cancer-related premature mortality cost an estimated €17 billion in 2008 [7]. In the United States, annual lung cancer treatment expenditures were estimated at $13.1 billion USD in 2014 and lost productivity due to premature lung cancer deaths was an additional estimated $36.1 billion USD [2,8,9].

Rationale
Systemic therapy can provide a meaningful clinical benefit for patients with advanced NSCLC, and several new therapies have been approved since 2010 [10]. However, there is a paucity of published information describing how these therapies are used in real-world clinical practice, especially for second and later lines of therapy. Increased understanding of how these treatments are used in routine clinical practice and the associated clinical outcomes may provide insights into how these therapies benefit patients, inform strategies that support development of new therapies, and ultimately decrease the global economic and societal burden of NSCLC.

Objectives
The primary objective of this study was to describe treatment patterns and survival outcomes among patients who received second-line treatment for advanced NSCLC in routine clinical practice. The study also describes treatment patterns and available information on survival for the subset of patients who received third-line treatment.

Approach
To achieve these objectives, we conducted a systematic literature review and qualitative evidence synthesis of observational studies.

Literature search strategy
This systematic review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [11]. A review protocol does not exist. A tiered search string that included a combination of keywords and medical subject headings (MeSH; S1 Table) was used to search the following databases: BIOSIS Previews, Current Contents Search, Embase, Gale Group PROMT, International Pharmaceutical Abstracts, Medline, and SciSearch.

Study selection criteria
Observational cohort studies that included detailed information on second-and third-line treatment patterns in patients with advanced NSCLC were eligible for inclusion. Only Englishlanguage studies published in peer-review journals between January 1, 2010, and March 1, decision to publish, and preparation of the manuscript, but did not play a role in the study design, data collection and analysis. The specific roles of these authors are articulated in the 'author contributions' section.
2017, were considered. This date range was intended to capture current real-world data on advanced NSCLC treatment patterns in order to account for changes in treatment guidelines related to newly approved therapies and biomarker-guided treatment decisions. Studies that described only one type of therapy, case reports, clinical trials, conference abstracts, and Delphi panels were excluded. Additionally, studies that introduced selection bias with patient selection criteria like requiring platinum-based chemotherapy or a specific mutation were excluded since the intent of the qualitative review was to describe the treatment patterns in a broad, unselected patient population.

Study selection process
Article titles and abstracts were initially screened by one reviewer (JD) to identify articles that potentially fulfilled the study selection criteria. Full-text articles were independently evaluated for inclusion by 2 reviewers (JD and MM). Additional publications were identified through examination of references cited by the included publications and were included if they also fulfilled the selection criteria.

Data extraction
Two independent reviewers extracted key information from each publication, including study location, time period, methods, patient and tumor characteristics, treatment regimens by line of therapy, survival outcomes, response rates, and biomarker testing information. If multiple publications described a single study, the extracted data were combined, and each publication was referenced to reflect this circumstance. Discrepancies during study data extraction were resolved by consensus between the reviewers with reference to the articles. If data differed between 2 publications describing the same study, data from the more recent publication were used.

Main study measures
The main study measures included the proportion of treatment regimens used in secondand third-line advanced NSCLC treatment and overall treatment rates in first-, second-, and third-line therapy. Treatment lines were defined based on the reporting authors' definitions. Single-agent chemotherapies included systemic anti-cancer therapies other than targeted therapies that were prescribed as monotherapy. Combination chemotherapies comprised 2 or more systemic anti-cancer therapies, including both chemotherapies and targeted therapies. Single-agent targeted therapies included any targeted therapy prescribed as monotherapy. Investigational drugs included any drugs administered as part of a clinical trial. In studies that did not explicitly report treatment proportions among patients receiving second-or third-line treatment, the proportion was estimated based on the number of patients receiving a specific treatment regimen and the total number of patients receiving second-or third-line treatment. Survival outcomes were reported based on the original definitions described in each study.

Assessment of bias
Risk of bias and methodological quality of the studies were assessed using the 2013 RTI framework created for the Agency for Healthcare Research and Quality for observational studies [12]. Two authors independently performed this assessment (S2 Table).

Results
The literature search yielded 1,329 citations, and 10 additional citations were identified through examination of references cited by the included publications (Fig 1). After applying the selection criteria, 15 articles describing 12 different study cohorts were included in the qualitative data synthesis. Of the 12 study cohorts, 7 were retrospective medical record reviews or database analyses and 5 were prospective single-center or multicenter cohorts ( Table 1). The study cohorts included patients from Europe (7), North America (3), Asia (1) and South America (1). In general, there was minimal risk of bias within the studies, although several studies had issues that could lead to confounding of the reported outcomes, either due to the length of time during which the study was conducted or missing information on key study variables such as histology or use of oral therapies. Overall, the study results were judged as believable after accounting for limitations.

Patient characteristics
Median age at advanced NSCLC diagnosis ranged from 59 to 68 years in 10 studies that reported this information (Table 1) [13][14][15][16][17][18][19][20][21][22][23][24][25]. Males comprised a larger proportion of the study population in all cohorts. There were 2 studies that included only patients with a single NSCLC histology-one with only squamous patients, and one with only non-squamous patients [26,27]. Among the other studies, adenocarcinoma was the most common NSCLC histology, with the proportion of adenocarcinoma patients ranging from 33% in Brazil to 77% in Japan [13,23]. Two studies enrolled only stage IV patients [20,23]. The proportion of patients who never smoked was highest in the Japanese cohort [13,]. Five studies reported performance status after first-line therapy or at initiation of second-line therapy, with most patients having a Karnofsky Performance Status of 70 or better or an Eastern Cooperative Oncology Group score of 0 or 1 [15,16,18,21,24,27].

Treatment patterns
Second-line treatment. Fig 2 and Table 2 describe the distribution of second-line treatment regimens in each of the studies. The proportion of patients receiving second-line therapy among the studies varied depending on how the study cohort was selected and what treatments were included. In studies that followed patients from initial NSCLC diagnosis, the proportion of patients who received second-line treatment ranged from 8% in a population-based Canadian study that did not include oral therapies (epidermal growth factor receptor [EGFR] tyrosine kinase inhibitors [TKIs]) to 53% in a German study at a single institution [20,24,25].
Pemetrexed was one of the most commonly prescribed second-line treatments in the studies that included patients treated after its initial approval in 2004 [15-18, 22, 24, 25, 27]. The study conducted in Canada reported that pemetrexed use increased to 70% among treated second-line patients in 2008, when government funding was approved for pemetrexed [20].
Second-line treatment with EGFR-TKIs increased in the European studies after initial approval of erlotinib in 2005 [15-19, 21, 22, 24, 25]. This uptake occurred prior to the  widespread introduction of EGFR mutation testing, which was not required at the time of initial erlotinib approval. Platinum-based doublet chemotherapy was administered in all 7 studies that described the use of second-line combination chemotherapy regimens [13, 15-17, 19, 21, 24-26]. The most frequent use of combination chemotherapy was reported in a study that included only patients with squamous histology [26].
Third-line treatment. Nine of the 12 studies reported detailed information on patients who received third-line therapy [15][16][17][18][19][21][22][23][24][25]. In these studies, approximately 30% of patients received third-line treatment (S1 Table). Single-agent chemotherapy accounted for about 50% of third-line treatments in general, and targeted therapy accounted for almost 40% [15-18, 21, 22, 24, 25]. Docetaxel, gemcitabine, vinorelbine, and pemetrexed were the most Patterns in second-and third-line treatment for advanced NSCLC Patterns in second-and third-line treatment for advanced NSCLC frequently administered third-line chemotherapies. Erlotinib was the most common singleagent targeted therapy in the third-line setting in all countries.
Other lines of therapy. All 12 studies reported details about the distribution of first-line treatment regimens. The most frequently administered first-line treatment across all countries was platinum-doublet chemotherapy (S3 Table). Three European studies reported use of targeted therapy in the first-line setting. In an Italian cohort, gefitinib, bevacizumab, and erlotinib were administered to 6%, 4%, and 1% of first-line treated patients, respectively [15,16]. Erlotinib was administered to 3% of first-line patients in a German cohort [24,25]. Only 4 patients (0.4%) in a French cohort received an EGFR-TKI in first-line therapy [21].
In Japan, fourth-line chemotherapy was reported in 17.7% of the patients who received first-line chemotherapy. The top 3 anti-cancer treatment regimens in the fourth-line setting were single-agent S-1 (22%), docetaxel (21%), and gefitinib (21%) [13]. In Brazil, 2.5% of all patients with stage IV NSCLC (4.3% of the patients who received first-line chemotherapy) received fourth-line chemotherapy [23].
In 5 studies that described median OS from time of second-line treatment initiation in routine clinical practice, median OS ranged from 4.6 months (95% CI, 3.8-5.7) to 12.8 months (95% CI, 10.7-14.5) [13,19,22,24,25,27]. Median OS estimates were close to 8 months in 2 studies that reported this information specifically for patients with adenocarcinoma in the second-line setting [18,22,27]. Median OS estimates were reported from the time of third-line treatment initiation in 4 studies and median OS ranged from 3.8 (95% CI, 2.6-5.4) to 12.0 months (95% CI, 9.3-14.2) [13,24,25] Biomarkers Two studies reported biomarker testing and treatments prescribed based on biomarker status, one in Italy (LIFE) and one in the US that included only patients with adenocarcinoma (Table 4) [15,16,27]. In the US study, EGFR testing frequency increased significantly from 2.3% between 2007 and 2009 to 32% in the first 6 months of 2011 [27]. The LIFE study, which was conducted in 2011-2012, reported that 60% of Italian patients received biomarker testing [15,16]. In this study, biomarker testing was more frequent among patients who were younger, female, never-smokers, and those with adenocarcinomas.
Both studies provided evidence of biomarker-driven therapy choices within their cohorts. The LIFE study reported that 26 of 37 (70%) patients who received EGFR-TKIs as first-line therapy were known to have EGFR-activating mutations prior to first-line treatment [15,16]. In the US cohort, 50% of 24 patients with known EGFR mutations received erlotinib as second-line treatment compared with 17% of the 89 EGFR wild-type patients. The use of erlotinib was significantly more common in patients with EGFR mutations compared with EGFR-wild type patients (P < .001) [27]. Patterns in second-and third-line treatment for advanced NSCLC

Discussion
The objective of this study was to describe real-world treatment patterns and survival outcomes for patients with advanced NSCLC who received second-line or later treatments, through a review of recently published observational studies. To our knowledge, there are no systematic reviews that summarize this information. By qualitatively reviewing this information from multiple countries, we provide a broad overview of how patients are managed in this setting around the world.
Overall, the retrieved studies showed how newly approved therapies are rapidly integrated into clinical practice, and how clinical practice evolves in response to therapeutic advances. In addition, most studies demonstrated a high level of clinician adherence to international treatment guidelines, such as those from European Society of Medical Oncology and the US National Comprehensive Cancer Network [28,29]. Among the studies that focused solely on patients with non-squamous NSCLC, single-agent pemetrexed, docetaxel, or EGFR-TKIs were the most common treatment choices, which aligns with current guidelines. However, there were examples of non-adherence to guidelines, namely use of platinum-doublet chemotherapies as second-line treatment. This was highest in the allsquamous cohort, which is consistent with the limited number of approved second-line treatments for this particular subgroup of patients with advanced NSCLC that were available at the time these studies were conducted [26]. Moreover, while specific patient-level information was not available from these studies, it is possible that use of platinum-doublet chemotherapy in the second-line setting could be motivated by oncologists' perceptions of the superiority of these particular treatment regimens.
Survival after initiation of second-line treatment was reported in most studies, and was generally consistent with outcomes reported from pemetrexed, docetaxel, and erlotinib clinical trials [30][31][32][33]. The consistency of survival outcomes suggests that clinical trial results may be generalizable to real-world patient populations. Additionally, the data confirm the limited survival benefit of treatments that were available at the time these studies were conducted. An exception to this was the lower survival among second-line patients in Germany reported by Zietemann and Duell compared with outcomes reported from other European studies with similar outcome definitions [19,22]. No explicit reason for the lower survival reported in this study was identified based on patient and tumor characteristics, or the study setting. The range of survival outcomes reported by the studies included in this review exemplifies the variation that can occur across multiple studies and may be due to individual-level patient differences. In addition, these survival differences highlight the importance of examining population-level data and outcomes from multiple observational studies to understand the range of outcomes experienced in routine clinical practice and the limited ability to draw conclusions from studies that do not report detailed patient-level data.
In most studies that reported the proportion of patients with advanced NSCLC treated across lines of therapy, between one-third to one-half of those who received first-line treatment also received second-line therapy [14,16,20,21,[23][24][25][26]. None of these studies provided information to explain why patients did not receive later lines of therapy. However, Gridelli et al reported that poor performance status and older age were the most common reasons patients did not receive first-line treatment [17]. These reasons, in addition to a poor response to first-line treatment, may be why patients with advanced NSCLC do not receive later lines of therapy. Newer treatments such as immunotherapies may offer these patients an opportunity to receive treatment because of their more favorable toxicity profiles compared with existing treatments [34,35]. Additional studies that examine why patients are not treated could help identify opportunities to increase the proportion of patients with advanced NSCLC who derive benefit from treatment.
There was limited information reported on biomarker-guided therapy decisions in the studies included in this review, likely due to the time periods covered by the studies. In the studies covering more recent time periods, the rapid uptake seen in the use of targeted therapies and biomarker testing is encouraging. It reveals that biomarker-driven treatment decisions are being integrated into routine clinical practice. This information provides a positive outlook for the uptake of newer targeted therapies in the second-and third-line settings. The rapid uptake of anaplastic lymphoma kinase (ALK) inhibitors such as crizotinib and ceritinib in patients with advanced NSCLC harboring ALK mutations provides further evidence that biomarker-directed therapy is an important treatment option for these patients [36].
Biomarker testing may impact uptake of newly approved targeted therapies. More information on how and when biomarker testing is performed in routine clinical practice is needed, since it is unclear in the included studies whether biomarker testing was more commonly performed prior to first-line versus second-line treatment. Across the European studies, erlotinib use increased despite approval of EGFR mutation testing in Europe after erlotinib approval. This suggests a trend among clinicians of using new therapies when clinically appropriate, despite limitations in access to biomarker testing. These trends imply positive uptake of new immunotherapies like programmed death-ligand 1/programmed death 1 (PD-L1/PD-1) checkpoint inhibitors despite evolving biomarker testing guidelines.
A few limitations should be considered when interpreting the information provided in this review. A meta-analysis could not be conducted due to the heterogeneity of the study designs and data presentation in the included studies. There were substantial differences in patient selection procedures, data collection methods, and outcome measure definitions. Thus, a qualitative data synthesis was the most appropriate way to summarize these studies [37]. Moreover, the review was robust and systematic, which renders the search and presentation of results transparent and reproducible. Additionally, publication bias could influence the evidence presented in this review; however, studies were found from multiple countries. Finally, conclusions from the results presented herein are limited by the observational design of the included studies.

Conclusions
Real-world studies of second-and third-line treatment patterns in advanced NSCLC provide insights into how evidence from clinical trials impacts clinical practice. The studies included in this qualitative systematic review demonstrate the limited clinical benefit of the second-and third-line treatments for advanced NSCLC that were available at the time these studies were conducted. Within recent years, several novel therapies have been approved for use in these settings, and it will be important to determine the benefit that these and other new treatments may provide to patients with advanced NSCLC.
Supporting information S1