Real-World Utilization of Target- and Immunotherapies for Lung Cancer: A Scoping Review of Studies Based on Routinely Collected Electronic Healthcare Data

Routinely collected electronic healthcare data (rcEHD) have a tremendous potential for enriching pre-marketing evidence on target- and immunotherapies used to treat lung cancer (LC). A scoping review was performed to provide a structured overview of available rcEHD-based studies on this topic and to support the execution of future research by facilitating access to pertinent literature both for study design and benchmarking. Eligible studies published between 2016 and 2020 in PubMed and ISI Web of Science were searched. Data source and study characteristics, as well as evidence on drug utilization and survival were extracted. Thirty-two studies were included. Twenty-six studies used North American data, while three used European data only. Thirteen studies linked ≥1 data source types among administrative/claims data, cancer registries and medical/health records. Twenty-nine studies retrieved cancer-related information from medical records/cancer registries and 31 studies retrieved information on drug utilization or survival from medical records or administrative/claim data. Most part of studies concerned non-small-cell-LC patients (29 out of 32) while none focused on small-cell-LC. Study cohorts ranged between 85 to 81,983 patients. Only two studies described first-line utilization of immunotherapies. Results from this review will serve as a starting point for the execution of future rcEHD-based studies on innovative LC pharmacotherapies.


Introduction
Lung cancer is the most commonly diagnosed cancer worldwide (2.09 million cases in 2018). It accounts for 14.5% of the total cases of cancer in men and 8.4% in women, being the leading cause of cancer death in men (22.0%) [1,2].
Treatment of lung cancer relies on one or more therapeutic approaches among surgery, radiation therapy and pharmacotherapy [7]. Currently, a wide range of medications is available for the treatment of advanced NSCLC. The choice of a specific pharmacological regimen is mainly based on the stage of the cancer, although other factors such as the overall patient's health and lung function, as well as some specific molecular traits of the cancer itself, are also important. Early-stage NSCLC shows no overt clinical symptoms, and surgery represents the treatment of choice. In such cases, pharmacotherapy can be used both before, as neoadjuvant treatment aimed to reduce the size of the tumor, and after surgery, as adjuvant treatment intended to decrease the risk of cancer recurrence [5,8]. In advanced stages, where cancer has already spread, treatment choice depends on the specific site and number of metastases, other than age and overall health status of the patient. In particular, while pharmacotherapy of SCLC is based mainly on standard chemotherapy (the FDA approved nivolumab in August 2018), during the last 15 years, the pharmacological treatment for advanced stage NSCLC was revolutionized by the authorization of innovative anticancer therapies, such as target therapy and immunotherapy [9].
Target therapy includes drugs that can counteract specific mechanisms underlying the development of tumors [10]. These include the neutralizing antibody bevacizumab, which acts by binding the pro-angiogenic vascular endothelial growth factor (VEGF), the tyrosine kinase inhibitors (TKIs) directed to the epithelial growth factor receptor (EGFR), and TKIs of anaplastic lymphoma (ALK). Immunotherapy includes nivolumab (approved in 2015) and pembrolizumab (approved in 2016): these drugs inhibit the binding between lymphocyte protein death 1 (PD-1) and tumor ligand of PD-1 (PD-L1) by maintaining the immune system's response to the tumor [11,12]. In advanced NSCLC and in non-operable patients, some of these drugs are the first-line treatment (e.g., anti-TKIs) in patients with activating mutations in EGFR or ALK genes, while others are licensed as second-line treatment (e.g., Nivolumab) [8,[13][14][15]. No target therapies are approved for the treatment of advanced SCLC.
Knowledge on efficacy and safety of authorized anticancer drugs mostly relies on evidence from clinical trials [16]. Such studies are usually based on relatively small samples of strictly selected, well monitored, patient populations, which are generally followed for short time periods [17].
In this context, observational studies based on large databases of routinely collected electronic healthcare data (rcEHD) has the potential to complement information from clinical trials by allowing the observation of the "real world" clinical practice, thus leveraging data from wider and less strictly selected populations, during long-term follow-up periods [18,19]. Given also the hot topic of using big data, as well as artificial intelligence, for longitudinal data mining in healthcare [20], an overview of available data to conduct pharmacoepidemiologic studies is needed. Unfortunately, the conduction of such studies in the oncology setting remains often a challenge since information to reliably describe utilization of cancer drugs, patients' characteristics and outcomes are often scattered in distinct data sources.
We performed a scoping review [21] of the published rcEHD-based studies concerning the utilization of target-and immuno-therapies in LC patients with the aim of providing a structured overview of the available studies to facilitate the design and benchmark of future works on this topic.

Literature Search
We searched PubMed and ISI Web of Science databases for retrieving the articles of interest that were published from January 2016 to August 2020. Due to the approval of immunotherapy in 2015, January 2016 was chosen as starting date to give a more upto-date picture of the issue [11]. The search string used was composed by three sets of keywords respectively related to the concepts "lung cancer", "drug-utilization measures", and "type of study/Data", respectively (see Supplementary Materials-Table S1 for more details). Snowballing search was also conducted to retrieve additional papers of interest by examining the references cited in the included articles.

Eligibility Criteria
Retrospective observational studies based on information retrieved from rcEHD that reported evidence on target therapies and immunotherapies in patients with lung cancer were selected. Eligible studies had to be published between January 2016 and August 2020 and written in English. Studies based on ad hoc data collection or with no abstract or full-text available were excluded.

Study Selection
Two authors (AS and GH) screened all titles and abstracts of the references retrieved. Potentially relevant studies were further assessed through examination of full texts. The reviewers worked independently, in parallel, and blinded to each other. Disagreements between the two reviewers were solved through discussion with a third author (GR).

Data Extraction
The following information was extracted from the included studies: (i) Data source characteristics: type of source, name, catchment area. Notably, data source types were classified into three main categories: (a) administrative/claims data (i.e., data for health system planning and management and health assistance claims), (b) "medical/health records" (i.e., documentation of clinical care) and (c) "cancer registries" [22,23]; (ii) Study characteristics: study population, population size, cohort type (populationbased, hospital-based), study period, follow-up duration and drugs or drugs regimens. Additionally, relevant information items such, as cancer-related characteristics, patients-related characteristics, drug utilizations, vital status, were also classified by sources of rcEHD used, whenever possible; (iii) Information on the utilization of target-and immunotherapies based on treatment line and LC histology (e.g., pattern of use, frequency molecular testing, survival).
As for study selection, two authors extracted independently the data (AS and GH), and a third author (GR) was consulted in case of disagreement.
In particular, median overall survival (OS) values were extracted, whenever reported. Median OS values were grouped by treatment line and presented as the range between the maximum and minimum reported value.

Literature Search Results
A total of 594 study references were retrieved from PubMed and ISI Web of Science ( Figure 1).
Screening of titles and abstracts allowed the selection of 131 potentially eligible studies. Among them, a total of 32 studies fulfilled the eligibility criteria and were finally included into the review . No further studies were retrieved through a snowballing search.

Sources of rcEHD Used for Information Retrieval
Medical/health records and cancer registries were most frequently used to retrieve cancer-related information (i.e., histology, stage, molecular/genetic characterization, tumor response and disease progression-see Table S2 Supplementary Materials): on a total of 23 studies where the source used to retrieve the reported cancer-related information could be assessed, 13 used medical/health records [25,29,33,37,41,43,[47][48][49][51][52][53][54] and seven used cancer registry data [26][27][28]32,36,42,50]. Notably, tumor response was reported in four studies only: the information was always retrieved from medical/health records [25,33,43,48]. In three studies based on administrative/claim data, instead, proxies of cancer-related information were used to identify tumor histology and/or stage [31,39,55]. A study based on French administrative healthcare data used bevacizumab or pemetrexed dispensing as a proxy for non-squamous NSCLC histology [55]. Two other studies based on administrative/claims data from US, identified patients with metastatic cancer by using algorithms based on a combination of ICD-9CM codes (e.g., excluding patients with a claim for lung surgery, and then selected only those patients with ICD-9CM codes referring to a metastatic disease-see Table S3 Supplementary Materials further details on algorithms used to derive missing variables from administrative/claim data) [31,39].
The use of first-line immunotherapy was described by two studies [52,54]. Information about pembrolizumab and nivolumab (e.g., changes in treatment line during study period and trend for utilization) was reported in both studies. The study of Molife et al., reported also that no patients received atezolizumab as a first-line treatment in a population extracted from the US Flatiron healthcare database from 2014 to 2017 [54].
Eight studies from the US concerned anti-EGFR therapies as second-line in patients with advanced NSCLC [34,37,[49][50][51][52][53][54]. All the eight studies described the use of erlotinib, of which three described also the use of gefinitib, afatinib and osimertinib [49,53,54]. Five studies showed that the incidence of use of second-line anti-EGFR utilization among advanced NSCLC patients ranged between 3.6% and 18.6% [34,37,50,51,53]. Four US studies reported the use of anti-ALK therapies as second-line therapy [46,48,52,54]. Two out of four studies described the use of anti-ALK medications in an ALK mutated NSCLC cohort [46,48], while the remaining two studies concerned a cohort of patients included regardless of molecular characteristics of the tumor.

Utilization of Target-or Immuno-Therapies for Neuroendocrine Lung Cancer
One out of the 32 included studies referred to patients with neuroendocrine lung cancer [35]. Using the MarketScan Database and PharMetrics Database between July 2009 and June 2014, the authors reported that in a total of 785 patients, 78.2% started first-line therapy with cytotoxic chemotherapy, 18.1% with somatostatin analogues, and 1.1% with other drugs such as sunitinib or everolimus.

Utilization of Target-or Immuno-Therapies for Unspecified Lung Cancer Histology
Using administrative data only, two studies (two from the US) included advanced stage lung cancer patients regardless whether they were diagnosed with NSCLC or SCLC [31,39,55]. Both studies used data sources from the USA [31,39]: the first described the first-line use of biologic therapy (bevacizumab, crizotinib, erlotinib and cetuximab) in patients with metastatic lung cancer by site-of-care [31] and the second one the use of erlotinib in patients with EGFR mutated metastatic lung cancer [39].

Discussion
With this scoping review we provided a structured overview of the available literature concerning recently published rcEHD-based studies concerning the utilization of target-or immunotherapies for LC. Our results highlighted a paucity of studies performed in Europe concerning immunotherapies, particularly as first-line pharmacotherapy, and the absence of papers reporting on the utilization of innovative drugs in SCLC patients. Focusing on the different types of rcEHD and methodologies used to retrieve information, results from this review represent a starting point for future studies on this topic, also highlighting current gaps of knowledge and facilitating access to pertinent literature both for study design and for benchmarking of results.
As for countries of data provenance, most of the studies included in this scoping review were conducted using data from the USA or Canada. This is probably because in regions other than North America, healthcare data are often scattered in different and heterogeneous databases, so that the performing studies on rare events that requires information from different healthcare settings, as in the case of lung cancer, remains a challenge [56]. Moreover, the approval of new anticancer medication in Europe is often delayed compared to the US [57], possibly contributing to the higher number of studies from the US included in this review compared to those using European Union (EU) data.
Indeed, results from this review demonstrated that a unique source of electronic healthcare data among administrative/claims data, medical/health records, cancer registry is often insufficient for performing an observational study on the real-world utilization of drugs for LC, as well as for other types of tumor [58].
Concerning the specific sources of rcEHD used for information retrieval, medical/health records were the most frequently used source of information for assessing drug exposure. Among the studies included in this review, administrative/claims data were less frequently used for retrieving such information. This was probably because in-hospital drug utilization might not always be tracked at patient-level in this type of rcEHD [30][31][32][33][34]36,39,55]. Also information on treatment-line is not usually available in administrative/claims data, although ad hoc algorithms can be adopted to derive this information (Table S3 Supplementary Materials) [31,32,35,55]. Moreover, administrative/claims data usually do not record clinical information, such as tumor stage, histology, or gene mutations, which are crucial for studying drug utilization patterns and health outcomes in cancer patients. In this respect, the use of medical/health records or cancer registries appeared to be in most cases necessary [25,28,29,32]. Our results showed that information on disease progression and tumor response was only retrieved from medical/health records [25], while vital status was assessed using administrative/claims data or medical health records, although the former are usually considered as the gold standard for such information [59]. Indeed, each type of data source has its strengths and limitations with respect to the specific research question that needs to be addressed. Even within each of the three general categories of data sources adopted in this review [19], a significant heterogeneity in terms of content and validity has to be expected (see Table S2 Supplementary Materials). Therefore, as has already happened in other contexts [56,60,61], fostering the development of methodologies for leveraging data diversity will be of paramount importance for the generation of solid evidence on the real-world utilization of drugs in LC.
As for evidence on the real-world utilization of innovative anticancer drugs, most of the included studies concerned patients with advanced stage NSCLC while no studies focusing on SCLC were found. The absence of licensed target therapies and the recent approval of immunotherapies for SCLC (Nivolumab was the first approved in August 2018 in the US [62]) apparently explains the absence of any published study focusing on SCLC in our literature review. Given the very low prevalence of SCLC [8], rcEHD has the potential to play an important role in capturing and studying far larger populations of SCLC patients than those recruited in clinical trials. The orphan designation of different drugs intended for the treatment of SCLC has promoted the study of a number of promising treatments [63], mainly immunotherapies, that were recently marketed, or will be possibly approved in the near future [62,64]. However, further initiatives are desirable to foster SCLC genotyping for the discovery of new molecular targets useful to develop innovative medications. Findings from this review showed that available evidence on immunotherapies from rcEHD-based studies concerning immunotherapy used for advanced stage NSCLC is still scarce, particularly with respect to their use as first-line pharmacotherapy. Notably, only two studies reported evidence on the real-world utilization of immunotherapies administered as firstline pharmacotherapy in patients with advanced NSCLC [52,54]. Such paucity of literature is mostly due to the recent approval of this class of anticancer medications for such indications. In fact, pembrolizumab was the first immunotherapy approved for first-line treatment of advanced NSCLC in 2017 [11,12]. The reported estimates of the incidence of use of immunotherapies as second-line pharmacotherapy for NSCLC, instead, appeared extremely variable from one study to the other mostly due to the different study period, cohort characteristics, and active principles concerned (from 9.8% to 48.8%) [29,34,41,53]. Such heterogeneity of study characteristics and results, however, represents an important resource for benchmarking results of future studies. Conversely, from immunotherapy a markedly higher number of studies on target therapies as first or second-line treatment for NSCLC were found. These studies provided information on the real-world utilization of such a class of medications, such as estimates of the frequency of the use in the relevant study populations, by histology as well as by molecular test execution.
The main strength of this review is the systematic approach adopted for reviewing the available body of recently published literature on the topic, with an in-depth screening of the records retrieved from two comprehensive databases like PubMed and ISI web of science. In particular, the choice of including studies published starting from 2016 was mainly due to the recent approval of some of the drugs and indications of interest (e.g., the first included studies concerning immunotherapies was published in 2017). Moreover, this approach, other than increasing the efficiency of the literature search efforts (i.e., the number of observational studies published increased in the last few years [65]), allowed us to provide an overview of studies concerning the most up-to-date evidence and methodologies on the topic. Indeed, given the scoping nature of this review, quality assessment of the included studies was not performed.

Conclusions
In conclusion, this scoping review provided a structured overview of the published rcEHD-based studies that investigated the real-world utilization of target and immunotherapies in lung cancer patients. The characteristics of studies included in this review showed that record-linkage of different sources of rcEHD often appears to be necessary. Cancerrelated information were mainly retrieved from medical/health records or cancer registries while information on drug utilization or vital status were extracted in most of cases from medical/health records or administrative/claim data. As for evidence collected on the utilization of innovative medications for lung cancer, our results highlighted a paucity of studies performed in Europe as well as concerning immunotherapies, particularly as first-line pharmacotherapy. Notably, no study reporting drug utilization evidence concerning SCLC patients was found due to the absence of licensed target therapies and the very recent approval of immunotherapies for this indication.
Finally, this work will serve as a starting point for the execution of future real-world studies based on rcEHD facilitating access to pertinent literature both for study design and for benchmarking of results.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ijerph18147679/s1, Table S1; Research Strategy, Table S2; Sources of the data used by the included studies per study variable, Table S3; Algorithms used to extract information from administrative/claims data as reported in the papers included in the review, Table S4; Minimum and maximum median OS reported per treatment-line among stage III-IV NSCLC patients.
Author Contributions: M.Z., R.G., G.R. and S.D. conceived the study. A.S., F.S. and G.R. analyzed data and wrote the manuscript; A.S., G.H., C.B. and G.R. selected eligible articles and extracted information from retrieved studies. P.R. contributed to the interpretation and discussion of study results. All authors revised and approved the final version of the paper. All authors have read and agreed to the published version of the manuscript.