Real-world treatment patterns in resectable (stages I–III) non-small-cell lung cancer: a systematic literature review

Lung cancer is the leading cause of cancer-related mortality, accounting for almost one in five of all cancer deaths [1]. The majority (80–85%) of lung cancer cases are non-small-cell cancer (NSCLC) [2], with approximately 50% of patients with NSCLC being diagnosed in the non-metastatic setting [3] and an estimated 20–25% being candidates for surgery [4]. However, prognosis remains poor, with a 5-year overall survival (OS) of 92% in stage IA1, decreasing to 13% in stage IIIC [5].

Clinical practice guideline recommendations for the treatment of resectable NSCLC are based on stage. Surgery is the mainstay of treatment in stages I–II, but it is also recommended in selected patients in stage III. In stages II–III, chemotherapy after surgery is recommended by the European Society for Medical Oncology (ESMO) and National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology (NCCN Guidelines^®) [6,7]. While ESMO guidelines specify that adjuvant chemotherapy is preferred over neoadjuvant chemotherapy, NCCN Guidelines^® state that neoadjuvant chemotherapy may be a treatment option among patients with stage IIIA (T3, N1) who are candidates for adjuvant chemotherapy. If and to what extent, clinicians and patients adopt these treatment modalities may be determined from published observational studies that capture real-world clinical practice.

The objective of this systematic literature review was to characterize the treatment patterns according to treatment modalities, including the proportion of patients receiving surgery, modality and treatment timing with respect to surgery (i.e., neoadjuvant, adjuvant and peri-operative chemotherapy [CT], chemoradiotherapy [CRT] or radiotherapy [RT]) and the use of specific CT regimens in the real-world treatment of resectable stages I–III NSCLC in North America, Asia and Europe and, in doing so, assess how aligned real-world treatment practice is to published treatment guidelines.

Materials & methods

This systematic literature review was conducted according to the Cochrane collaboration methodology and reported according to Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidance [8,9]. The PRISMA checklist is available in Supplementary Table 1.

Search strategy

Two electronic databases (MEDLINE, EMBASE) were searched on 6 May 2020; abstracts from major oncology and lung cancer conferences were searched either electronically or manually in the previous 2 years. The reference lists of all included articles were hand screened for any additional relevant publications. Authors were contacted when additional information was needed. The search strategies are provided in the supplementary materials (Supplementary Tables 2 & 3).

Eligibility criteria & study selection

Eligibility criteria were pre-specified according to the patient, intervention, comparator and study design (PICOS) framework for evidence-based reviews. Eligible studies included adult patients with resectable (stages I–III) NSCLC in which a breakdown by therapy (e.g., surgery with or without neoadjuvant, adjuvant or peri-operative therapy) was provided. Eligible studies were observational and multi-centered and were conducted after 2000 in Canada, China, France, Germany, Italy, Japan, South Korea, Spain, Taiwan, the UK or the USA. Studies reported in French, German, Italian and Spanish were translated to English and eligible for inclusion; studies in other languages were excluded. Finally, studies with a sample size of 50 or less were excluded (deemed less representative than larger studies). Citations were screened for inclusion by two reviewers; consensus was performed by a third independent reviewer to arbitrate discordant results.

Data abstraction

Data pertaining to study design, patient characteristics and treatment were abstracted. Data were abstracted by one extractor and validated against the primary sources by a second extractor. For each study, the proportion of patients receiving various treatments as well as the proportion of patients receiving further treatment after a first recurrence were calculated. Within-study and across-study temporal trends were abstracted where available.

Assessment of validity

The risk of bias was not assessed due to a lack of an appropriate tool for the type of studies identified in this review. Instead, the internal and external validity of each study's findings related to study population and data source were assessed for each study by two reviewers; conflicts were resolved by a third reviewer. The assessment of external validity was made based on generalizability of the population, sampling frame (including geography or type of practice) and temporal sampling frame (i.e., study period). The internal validity relating to the data source captured the extent and type of missing clinical data. The level of concern was assigned based on exclusions affecting generalizability and was categorized as a concern, some concern and no concern.

Treatment pattern description

Treatment patterns were described within three populations: all patients with stages I–III NSCLC; a subgroup with resected stages I–III NSCLC; and a subgroup of resected patients receiving CT/CRT.

Within studies capturing a full population (i.e., not restricted to resected NSCLC), the proportion of patients receiving surgery was calculated and summarized for each study. Surgery was defined as a procedure in which the lung or part of the lung was surgically resected (thus omitting exploratory thoracotomy).

Within the resected population, treatment was described by three treatment modalities: surgery (with or without RT), neoadjuvant CT/CRT and adjuvant CT/CRT. Peri-operative CT/CRT was described separately on a study basis, since this was reported in only a few studies. Neoadjuvant therapy was defined as therapy administered prior to surgery. Adjuvant therapy was defined as therapy administered after surgery. Peri-operative therapy was defined as therapy administered both prior to and after surgery. Treatment was defined as bimodal when it included surgery and CT and defined as trimodal when it included surgery, CT and RT. The proportion of patients receiving each modality, among all resected patients, was calculated and presented for each study.

In terms of CT regimens, the proportion of patients receiving each regimen, among all resected or potentially resectable patients who received CT or CRT was summarized. Treatment patterns were presented by stage and country. The data were not pooled or meta-analyzed due to heterogeneous study designs, objectives and reporting.

Results

In total, 8107 records were screened for eligibility, 169 full-text records were reviewed and 20 publications from 20 unique studies were included in this review (Figure 1).

Of the 20 studies, ten were conducted in North America, four in Europe and six in Asia (Table 1). All studies were retrospective, except for the study by Pinquié et al., which was prospective with a retrospective component (to collect information on treatment administered prior to study start) [10].

Nine studies were conducted using databases or registries, three in North America [11–13], three in Europe [10,14,15] and three in Asia [16–18]. Seven regional studies were described as population based [19–25], with none at the national level. One study only included community care centers in southern states of the USA [26], two studies were small in size (<100 patients) [27,28] and one included only two hospitals [29].

Eligibility criteria varied across studies, particularly regarding the exclusion of certain patients (Table 1). These exclusions pertained to patients with specific types of surgeries (e.g., segmentectomy and wedge resection) [11,17,27], the extent of resections (e.g., incomplete resection R1 and R2) [12,15,17,18,28], treatment types (e.g., RT or surgery alone) [14,28] and modalities (e.g., neoadjuvant CT/CRT) [17,20,27]. These restrictions limited the authors' ability to make comparisons within similarly defined populations.

Disease stage varied across studies, with a majority of studies capturing stage III NSCLC [14,16,17,21,24,27]. Four studies included stage IV but provided separate reporting for stages I–III (Table 1).

Surgery

Of the nine studies that included resected and unresectable NSCLC (see Table 1), two large studies conducted in France and the USA allowed the proportions of resected stages I–III (stages IIIA and IIIB) NSCLC to be estimated at 50.3 and 38.1%, respectively (Table 2) [10,11].

The proportion of nonmetastatic patients receiving surgery tended to decrease with an increase in stage. In six studies that described treatment in stages I–III NSCLC among those receiving surgery, definitive RT or non-curative treatments (Table 2), the proportions of patients receiving surgery ranged from 68.2; to 78.6% in stage I [11,13], 45.5 to 45.8% in stage II [11,23] and 10.3% to 17.5% in stage III [11,19,21].

In four studies that described treatment among those treated with surgery or definitive RT (and thus excluded those who received no treatment or non-curative-intent treatment; Table 2), the proportions of patients receiving surgery were 94% in stage I [22] and 86.6% in stage II [22] and ranged from 20.8 to 59% in stage III [14,22,24].

Neoadjuvant, adjuvant, peri-operative therapies by stage

Stages I–III

Nine studies reported on the use of CT, CRT and RT in stages I–IIIA NSCLC [10–12,15,20,22,26,28,29]. Three studies conducted in France and the USA (two studies) were considered comparable and described the treatment patterns in the resected population (Table 3) [11,12,29]. Across those three studies, 9.2–24.6% received neoadjuvant CT/CRT prior to surgery, 24.2–28.3% received adjuvant CT/CRT following surgery, and 51.2–62.6% underwent surgery alone (with or without RT). The remaining six studies were not considered comparable due to missing modalities (i.e., neoadjuvant CT/CRT or surgery) and a lack of information on the timing of CT/CRT. Among those six, one study in Korea reported on the peri-operative setting, indicating that 7.8% received neoadjuvant CT, 9.8% received peri-operative CT and 82.4% received adjuvant CT [28].

Stage III

Thirteen studies reported on the use of CT, CRT and RT in stage III NSCLC [10,11,14–17,19–22,24,26,27]. Three studies conducted in Canada and the USA were comparable and described the treatment patterns in the resected population (Table 3) [11,21,24]. Arnold et al. and Moore et al. reported that 37 and 44% received neoadjuvant CT/CRT, 28.6 and 33.8% received adjuvant CT/CRT and 34.4 and 21.8% underwent surgery alone (with or without RT), respectively [11,24]. The study of Vinod et al. used the same data source but at an earlier time, and therefore it was superseded by Moore et al. [21,24]. The remaining ten studies were not considered comparable due to some modalities not being reported (i.e., neoadjuvant therapy or surgery) [14–17,20,26,27] or timing not being reported [10,19,22].

Stages I, II & III

Six studies described the treatment patterns in resected patients by stage (Table 3). The study of Arnold et al. was the only one that provided stage-specific proportions across stages I, II and III [11]. Within this study, the proportion of patients receiving neoadjuvant CT/CRT tended to increase with increasing stage, from 2.5% in stage I to 14.2% in stage II and 37.0% in stage III. The proportions using adjuvant therapy were 14.1% in stage I, 35.2% in stage II and 28.6% in stage III. Conversely, the proportion of patients receiving surgery alone (with or without RT) tended to decrease with increasing stage (83.4% in stage I, 50.6% in stage II and 34.4% in stage III). In the adjuvant setting, there tended to be an increase between stages I and II (from 14.1% to 35.2%) and then possibly a slight decrease in stage III (28.6%). These patterns were corroborated by across-study comparisons showing similar trends.

Peri-operative therapy was reported in three studies [14,21,28], with one study reporting that 0.8% of resected patients received peri-operative RT [21] and the other two studies reporting that between 10 and 52% of resected patients received peri-operative CT, CRT or RT. The two later studies, unlike the study of Vinod et al., excluded patients receiving surgery alone or with RT (see footnote of Table 1 for details of inclusion criteria) [14,28].

Additionally, the authors identified four studies that reported the proportions of patients receiving the different modalities irrespective of timing [10,16,17,19]. In the three most recent studies, the studies reported that about 30–33% of patients received surgery (with or without RT) in stage III resected NSCLC [10,16,19]. Treatment of the remaining patients was split between bimodal and trimodal therapy with no clear pattern (see Supplementary Table 4).

CT regimens

The CT regimens used alongside surgery are provided in Figure 2, with nine studies providing a breakdown by regimen in the adjuvant setting [10,15,17,18,20,25–28] and one study reporting separately in the neoadjuvant setting [10].

In one study in the USA, carboplatin-paclitaxel was the most common adjuvant regimen (30%), followed by cisplatin monotherapy (14%) [26]. In two studies in Canada, cisplatin-based regimens were more common, with some regional and/or temporal differences (e.g., cisplatin-gemcitabine was most common [80%] in British Columbia [2005–2010; stage II], whereas cisplatin-vinorelbine was most common [71%] in Ontario [2004–2006; stages 0–IV]) [20,25].

In two studies in Europe, platinum-vinorelbine was the most common regimen in the adjuvant setting (36% [15]–60% [10]). Chouaid et al. provided a breakdown by platinum agent; cisplatin-vinorelbine was the most commonly used treatment (ranging from 56% in Germany to 73% in the UK) [15]. In the neoadjuvant setting, Pinquié et al. reported that platinum-vinorelbine (31%), platinum-pemetrexed (21%) and platinum-gemcitabine (21%) were the three most common regimens but did not provide a further breakdown by platinum agent [10].

In Japan, a tegafur-uracil combination (UFT) was used in nearly all stage I patients (92.9%) in one adjuvant study [18] and in about a quarter of stage III patients (28.7%) in another adjuvant study [17]. Platinum-based CT was the most common CT in stage III in Japan (64.5–71.3%) [17,27]. In a study in South Korea, cisplatin-vinorelbine (39.8%), cisplatin-gemcitabine (15.1%) and carboplatin-vinorelbine (14.0%) were the most common adjuvant therapies [28].

Treatment sequencing after first recurrence

Information on treatment sequencing after a first recurrence was provided by four studies [10,15,23,26]. The study by Pinquié et al. was the only one to specifically report treatment modalities after a first recurrence [10]. Within the follow-up period of up to 2 years post-diagnosis, 31% of the resected population had gone on to receive subsequent treatment [10]. Among those receiving a second treatment, the most common modality for the second treatment was CT alone (45%), with only 7.8% of patients receiving a second surgery (alone or with CT, CRT or RT) [10]. There was no apparent major difference by stage (i.e., CT alone: 46% in stages 0–IIB vs 42% stage IIIA, second surgery [alone or with CT, CRT or RT]: 9.6% in stages 0–IIB vs 5.6% in stage IIIA) [10].

Two other studies reported on the proportions of the resected population receiving subsequent systemic therapy after a recurrence: 27% in Moore et al. (follow-up 10 years) and 50% in Chouaid et al. (follow-up 4.6 years) [15,23]. Finally, Buck et al. reported on specific treatment regimens, stating that carboplatin-paclitaxel, erlotinib and pemetrexed were the top three most common regimens after disease recurrence [26].

Temporal trends

Within-study temporal trends were available from two studies [12,29]. From 1995–2000 to 2001–2006, Riquet et al. reported an increase in the use of both neoadjuvant CT/CRT and adjuvant CT/CRT and a coinciding decrease in surgery (with or without RT) in stages I–III (neoadjuvant CT/CRT: 11.8–24.6%; adjuvant CT/CRT: 16.3% to 24.2%; surgery: 71.9% to 51.2%, respectively) [29]. Similarly from 2002 to 2011, Rajaram et al. reported increasing trends in the use of neoadjuvant and adjuvant CT/CRT and decreasing trends in surgery (with or without RT) use in stages II–III (neoadjuvant CT/CRT: 11.2–14.8%; adjuvant CT/CRT: 18–43.5%; surgery: 70.8–41.7%) [12].

Between-study temporal trends were available from two studies that used the British Columbia Cancer Agency database and had similar eligibility criteria; they showed that the proportion of patients receiving neoadjuvant CT/CRT was greater in the more recent study (2005–2012) compared with the older study (2000–2007) (i.e., 44.4 vs 13.6%, respectively) [21,24].

Assessment of validity

The validity assessment contained some level of concern about generalizability in each study (see Supplementary Table 5). There was concern about the generalizability of the eligible population in ten or more studies in each category assessed. In particular, the study period was found to be of some concern because all studies were at least 6 years old, and thus there might have been some evolution in treatment practices. There was no major concern about data sources; thus, information bias was assessed as being low overall.

Discussion

To the authors' knowledge, this is the first systematic literature review of the treatment patterns in resectable NSCLC in North America, Europe and Asia. There was a limited number of multicenter studies that reported the full breakdown of therapies, and within the available evidence, there was little information on the CT regimens used in the neoadjuvant setting.

Within the overall stages I–III NSCLC population that included both resected and unresectable patients, the authors observed that approximately 40–50% underwent surgical resection [10,11]. Surgical intervention was most common in stage I, with over two-thirds of patients receiving surgery, and was least frequent in stages IIIA/B, where it was received by around 10–20% of patients. This trend is perhaps unsurprising, given that one of the main considerations in selecting patients for surgical resection is lymph node involvement status, which by definition is absent in stages I and IIA and is present in TNM stage IIIB [6,7,30].

Within the surgically resected population of the USA- and Canada-based studies, adjuvant CT (with or without RT) was most commonly used in stages II and III, and neoadjuvant CT was most commonly used in stage III. In fact, in stage III disease, neoadjuvant CT usage with or without RT was often more common than an adjuvant modality, with temporal trends showing greater uptake of neoadjuvant CT and CRT use in stage III in more recent years [12,21,24]. These findings are generally consistent with the latest NCCN Guidelines^® recommending that adjuvant CT be used primarily in stages IIB–III and that patients who are candidates for adjuvant CT in stage IIIA (T1–3, N0–1) may be treated with neoadjuvant CT as an alternative option [6,7].

In Europe and Asia, there was less evidence in regard to neoadjuvant and adjuvant therapy use, although some information was available on treatment modalities (i.e., surgery and bimodal and trimodal treatment) [16,17]. The one European study that did report on neoadjuvant therapy reported increasing trends in its use, with a quarter of resected stage I–III patients receiving neoadjuvant RT/CT/CRT in 2001–2006 [29]; however, this was based on data from just two hospital centers in France, and therefore it was difficult to determine if this generalizes to other centers in France or elsewhere in Europe. Furthermore, it is unclear whether this trend of increased use continued, given the current recommendation by ESMO that aside from a potential benefit of downstaging via neoadjuvant therapy, adjuvant therapy is preferred over neoadjuvant therapy [7].

In terms of CT regimens, there were some notable differences between regions. Cisplatin was the most common platinum agent used in doublet combinations in Canada, France, Germany and the UK, in line with the ESMO guidelines stating that cisplatin is preferred [7]. However, in the USA, both carboplatin and cisplatin were used by a large proportion of the population [26] and in line with the NCCN Guidelines^® recommending both platinum agents. UFT use was exclusive to the Japan-based studies. Consistent with Japanese recommendations, UFT was the most common regimen used in stage I [18,31] and was evident in the treatment of stage III disease.

Despite the broad scope of this review and alignment with good practice guidelines, there are several limitations. First, the generalizability of the results may have been impacted by the absence of national population-based studies; although national studies were included, they tended to include mostly academic, teaching or specialized cancer centers, which have been associated with different patterns of care than community settings [12,32,33]. Applicability to current clinical practice may also have been limited due to study periods spanning the range from 2000 to 2014. However, temporal trends both within and between studies elucidated the evolution in treatment during that time, and since 2014 there have been no major approvals or changes in the guideline recommendations for the treatment of resectable NSCLC. In view of two novel immunotherapies showing positive results in the neoadjuvant and adjuvant settings after many years of limited innovation in this space [34,35], this review may help establish a ‘baseline’ from which the treatment of resectable NSCLC will evolve. Finally, the eligibility criteria of certain studies (e.g., exclusion of certain systemic treatments, RT or surgery types) limited the authors' ability to obtain and compare the complete treatment patterns in resectable NSCLC across studies. Overall, the studies were heterogeneous and the treatment patterns in the neoadjuvant, adjuvant and peri-operative settings could only be determined in subsets of studies; however, these subsets are useful in understanding contemporary treatments and historic trends in the treatment of resectable NSCLC.

Conclusion

To our knowledge, this is the first systematic literature review describing the real-world treatment patterns in stages I–III resectable NSCLC. Although a limited number of studies were available and the studies were heterogeneous, the review suggested an increase in neoadjuvant and adjuvant CT use with increasing stage, generally in line with guidelines. Additional real-world studies will be needed to fill the gaps in our understanding of the relative use of neoadjuvant, adjuvant and peri-operative therapy as well as CT regimens used in resectable NSCLC.