An exploratory comparative analysis of tyrosine kinase inhibitors or docetaxel in second-line treatment of EGFR wild-type non-small-cell lung cancer : a retrospective real-world practice review at a single tertiary care centre

Background Treatment for advanced non-small-cell lung cancer (nsclc), especially in patients with wild-type EGFR, remains limited. Recently, erlotinib, a tyrosine kinase inhibitor (tki) targeting EGFR mutation, was approved as second-line treatment in EGFR wild-type nsclc. Despite evidence of better overall survival (os) with chemotherapy than with tki in second-line treatment, data on the use of tki in the real-life clinical setting remain limited. The present practice review of tki use for secondand third-line treatment in EGFR wild-type nsclc also compares clinical outcomes for tki and single-agent docetaxel as second-line treatment. Methods Our retrospective cohort study included patients with EGFR wild-type nsclc treated at the Jewish General Hospital (Montreal, QC) between 2003 and 2013. Patients received a tki (erlotinib or gefitinib) in the second and third line or docetaxel in the second line. For each group, we determined os, disease control rate, progression-free survival (pfs), and event-free survival (efs). Results The tki group included 145 patients, with 92 receiving second-line treatment. In the control group, 53 patients received docetaxel as second-line therapy. In the tki group, os was 6.0 months; pfs, 2.7 months; and efs, 3.0 months. Comparing second-line treatments, os was 5.3 and 5.0 months respectively (p = 0.88), pfs was 2.5 and 1.8 months respectively (p = 0.041), and efs was 3.0 and 1.7 months respectively (p = 0.009). Conclusions In our study cohort, second-line therapy for EGFR wild-type nsclc with tki (compared with docetaxel) was associated with statistically better pfs and efs and noninferior os. Those findings raise the question of whether efs should also be considered when choosing second-line treatment in this patient population.


BACKGROUND
Lung cancer remains the leading cause of cancer-related mortality worldwide, and non-small-cell lung cancer (nsclc) constitutes about 85% of all lung cancers 1 .In Canada alone, an estimated 25,000 new cases of lung cancer were diagnosed in 2013.The 5-year mortality rate for this disease remains 14% in men and 20% in women, reflecting only modest advances in anticancer therapy in the past few decades.
Historically, platinum-based chemotherapy was the standard of care in the first-line treatment of nsclc, regardless of histology and molecular subtype 2 .Recently, the recognition of distinct populations has changed that treatment paradigm: pemetrexed and bevacizumab are used in nonsquamous adenocarcinoma 3,4 , crizotinib has been approved in patients with EML4-ALK fusion gene 5 , and tyrosine kinase inhibitors (tkis) target disease with mutations in the EGFR gene 6,7 .
Despite the advent of more targeted therapies, prognosis in advanced nsclc remains poor, with only 30% -50% of patients being able to tolerate second-line therapy 8 .Important uncertainties persist with respect to the treatment of these patients, most of whom do not carry distinct mutations.Three agents are currently approved for second-line treatment of nsclc after progression on first-line therapy: pemetrexed, docetaxel, and erlotinib (the latter targeting the EGFR mutation).
The gene encoding the epidermal growth factor receptor (egfr) tyrosine kinase is somatically mutated in a substantial fraction of lung cancers, making it an ideal target anticancer therapy 9 .Most activating mutations occur in the tyrosine kinase domain.The most frequent of these are exon 19 deletions that eliminate four amino acids from the tyrosine kinase domain, and exon 21 missense mutations that substitute arginine for leucine at L834R.Those mutations account for about 90% of all EGFR mutations and are the target of first-generation tkis such as erlotinib and gefitinib 10 .
However, even in the absence of exon 19 deletion and L834R substitution mutations, erlotinib has been shown to have benefit (compared with best supportive care) in the second-line treatment of nsclc patients deemed ineligible for further cytotoxic chemotherapy 11 .That finding led in 2004 to the approval of erlotinib, a reversible tki targeting the EGFR mutation, for second-line treatment of EGFR wildtype nsclc.Several phase iii trials looking at erlotinib in the second-line treatment of nsclc supported that decision, although, until recently, none were powered to study EGFR wild-type patients specifically (Table i).
In July 2013, Garassino et al. 14 compared docetaxel with erlotinib in second-line treatment specifically targeting EGFR wild-type nsclc.Those authors demonstrated better overall survival (os) with docetaxel than with erlotinib [8.2 months vs. 5.4 months respectively; adjusted hazard ratio: 0.73; 95% confidence interval (ci): 0.53 to 1.00; p = 0.05], and better median progression-free survival (pfs: 2.9 months vs. 2.4 months respectively; adjusted hazard ratio: 0.71; 95% ci: 0.53 to 0.95; p = 0.02).Another phase iii trial by Kawagushi et al. 15 showed that pfs was better with docetaxel than with erlotinib (2.9 months vs. 1.3 months, p = 0.01) in the subset of EGFR wild-type tumours, although the difference did not translate into better os (10.1 months vs. 9.0 months, p = 0.91).Results from both trials suggested better outcomes with singleagent chemotherapy in second-line treatment, although the effect on os was less clear.
However, real-life data on the use of tki in the treatment of patients with EGFR wild-type nsclc who have progressed on first-line chemotherapy remain limited.Here, we present a practice review on the use of tki in second-and third-line EGFR wild-type nsclc at a single tertiary care centre, and we compare clinical outcomes for tki compared with singleagent docetaxel in the second-line treatment of nsclc.

Population
This retrospective cohort study used electronic database and medical chart review to collect data.All sequential patients who were diagnosed with locally advanced or metastatic EGFR wild-type nsclc between 2003 and 2013 at the Peter Brojde Lung Cancer Centre, Jewish General Hospital (Montreal, QC), and who received tki as second-or third-line therapy were included.As a comparison sample, a cohort of all tki-naïve patients who were diagnosed with locally advanced or metastatic EGFR wild-type nsclc between 2003 and 2013 and who were treated with singleagent docetaxel in the second line were also included.The study protocol was approved by the institutional review board at the Jewish General Hospital.

Primary Outcomes
The primary objective of the study was to review the reallife practice of tki use in the second and third line for patients with EGFR wild-type advanced or metastatic nsclc.Treatment patterns were characterized by time to start of tki from the date of diagnosis of advanced nsclc, duration of treatment, adverse events, treatment discontinuations, and subsequent line or lines of therapy.

Secondary Outcomes
Secondary outcomes included the disease control rate (dcr), pfs, event-free survival (efs), and os (Table ii).Assessment of the dcr was made using the Response Evaluation Criteria in Solid Tumors (version 1.1) at time of first radiography after treatment start.Response was categorized as disease control or progressive disease.Progressive disease was defined as at least a 20% increase in the sum of diameters of the target lesions or as the appearance of 1 or more new lesions and was measured by the first available computed tomography images after the start of treatment.Progression-free survival was defined as time elapsed from the time of tki or docetaxel treatment initiation to the time of radiologic progression.Event-free survival was defined as the length of time after treatment initiation that the patient remained free of the complications or events that the treatment was intended to prevent or delay.Events included clinical or radiologic progression (or both) leading to discontinuation of treatment, death from any cause, and discontinuation of the tki because of adverse events.For patients who stopped treatment at the time of radiologic progression, efs was equal to pfs.Although efs has not been validated as a clinical measure in randomized trials, we feel that the defined measure better reflects real-world practice, in which, despite radiologic progression, treatment might be continued because of ongoing clinical benefit.Overall survival was defined as the period from the time of treatment initiation to the time of death.Associations between key characteristics and outcomes, including treatment with tki and the rate of response on tki therapy, were explored.
We also investigated differences in survival, by demographic and clinical characteristics, for patients with locally advanced or metastatic EGFR wild-type nsclc in the study sample who received second-line tki treatment and for the representative sample of tki-naïve nsclc patients who received second-line docetaxel.

Statistical Analysis
Demographic and clinical data were summarized as means, standard deviations, and medians for continuous variables, and as numbers and percentages for categorical variables.All summarized data are presented separately for the tki-treated group and the tki-naïve group.
Clinical outcomes data are presented as the proportion of patients experiencing the outcomes of interest (disease control, progression, survival).Estimates of pfs, efs, and os (median, 95% ci) were generated.Kaplan-Meier methods for censored data were used to descriptively evaluate pfs, efs, and os.
An exploratory analysis investigated differences in os between the tki-treated patients and the tki-naïve group.For that comparative analysis, log-rank or Breslow statistics were used to make nonparametric comparisons of the Kaplan-Meier analysis (unadjusted analyses).
An exploratory analysis separately presents demographic and clinical data for the second-line tki-treated EGFR-negative group and the tki-naïve EGFR-negative group who received docetaxel in the second line.Differences in key characteristics between the groups were investigated.
The statistical analyses were conducted using IBM SPSS Statistics (version 20: IBM, Armonk, NY, U.S.A.), and statistical significance was accepted at p values of 0.05 or less.

RESULTS
Table iii presents the clinico-demographic characteristics of the 145 tki-treated patients included in the study.Most of the patients were ex-smokers or current smokers (73.8%) who had a good performance status at diagnosis (80.0%ecog 0-1).Of those patients, 79.3% presented initially with advanced-stage disease (stage iiib or iv).By the time

Overall survival
The interval from time of treatment initiation to time of death Disease control rate Determined using RECIST 1.1 at the time of first radiography evaluation after treatment onset (Disease control was defined as complete response, partial response, or stable disease.) Progressive disease At least a 20% increase in the sum of the diameters of target lesions or the appearance of 1 or more new lesions (measured on the first available computed tomography evaluation after start of treatment) Progression-free survival The interval from initiation of second-or third-line treatment to radiologic progression Event-free survival The duration of second-or third-line treatment (If treatment was discontinued at time of radiologic progression, event-free survival was then equal to progression-free survival.)

Event
Clinical or radiologic progression leading to discontinuation of treatment, death (from any cause), or discontinuation of tyrosine kinase inhibitor because of adverse events RECIST = Response Evaluation Criteria in Solid Tumors.
of tki or docetaxel initiation, all patients had advanced or metastatic disease.Adenocarcinoma and bronchoalveolar carcinoma were the primary histologies observed.
A tki was given to 92 patients in second-line treatment and to 53 patients in third-line treatment (Figure 1).Of the 92 patients who were treated with a tki in the second line, 26 went on to third-line chemotherapy, and of the 53 who received a tki in the third line, 20 received fourthline chemotherapy.
Median time from diagnosis of advanced disease to tki initiation was 7.4 months (95% ci: 6.84 to 8.02 months).Median time of efs (measured by the duration of tki treatment) was 3.0 months (95% ci: 2.6 to 3.47 months).First computed tomography imaging was performed at  the average of 3.5 months after initiation of treatment.The dcr in the tki group was 32.4% (32.6% in second-line treatment and 32.1% in third-line treatment, Table iv).Treatment with tki continued beyond progression in 28 patients: 16 in the second line (17.4%), and 12 in the third line (22.6%).Median os was 6.0 months (95% ci: 4.5 to 7.5 months; Figure 2).
Overall, tkis were well tolerated (Table v).Most patients discontinued treatment because of disease progression (84.0%), with a small proportion discontinuing treatment because of grade 3 or greater side effects (8.3%).At the time of study closure, 9 patients were still receiving tki treatment, of whom 7 were receiving tki in the second line, and 2 in the third line.

Comparison of Second Line TKI and Docetaxel
The number of non-smokers was statistically significantly greater in the tki group than in the docetaxel group (29.3% vs. 6.1%, p = 0.001).A good ecog performance status was also significantly more prevalent in the tki group than in the docetaxel group (Table iii).
Mean time to initiation of second-line treatment from progression of disease after first-line treatment was similar in both groups: 3.6 ± 5.8 months in the tki group and 4.6 ± 5.0 months in the docetaxel group.The dcr was better in the second-line tki group than in the docetaxel group (32.6% vs. 26.5%,Table iv), but the difference did not reach statistical significance.Progression-free survival was significantly better (p = 0.041) in the second-line tki group (2.5 months; 95% ci: 1.9 to 3.1 months) than in the docetaxel group (1.8 months; 95% ci: 1.7 to 2.0 months).Event-free survival was significantly better (p = 0.009) in the second-line tki group (3.0 months; 95% ci: 2.38 to 3.62 months) than in the docetaxel group (1.7 months; 95% ci: 1.57 to 1.89 months).Overall survival was similar in both groups: 5.3 months in the tki group (95% ci: 4.0 to 6.6 months) and 5.0 months in the docetaxel group (95% ci: 3.9 to 6.5 months).The difference did not reach statistical significance (log rank p = 0.88, Figure 2).

DISCUSSION
Before publication of the tailor study in July 2013, erlotinib had been used quite extensively in the clinical setting for second-line treatment of nsclc.However, there is a paucity of data concerning outcomes in patients with EGFR wildtype nsclc receiving a targeted agent, and even fewer data comparing outcomes with the use of tki or with singleagent chemotherapy.
Overall, patient demographics in the present study are comparable to those in previously published phase iii studies studying tki for the second-line treatment of advanced nsclc.Median age in the second-and third-line tki groups (63.6 years) is similar to that in previous pivotal trials 11,14,16 of second-line treatment in advanced nsclc (61.4,66, and 59 years respectively).The predominance of a good ecog performance status and of adenocarcinoma on histology examination is also reflective of other pivotal trials.
Our results seem to show that treatment with tki in the second-line setting was not inferior to docetaxel therapy.In the second-line tki group, pfs and efs were statistically significantly better.Trends toward better dcr and os were observed with tki (compared with docetaxel), although the differences were not statistically significant.
The os in our second-line tki cohort (5.3 months; 95% ci: 4.0 to 6.6 months) was similar to that reported in the tailor study (5.4 months; 95% ci: 4•5 to 6•8 months) and inferior to that reported in the delta study (9.0 months).However, in contrast to both the tailor and the delta studies, os in the docetaxel group was not superior to that in the tki group 11,12 (Table i).The better pfs seen with tki than   with docetaxel differs from the results of both the tailor and delta trials; the efs, which was also significantly better in the tki group than in the docetaxel group, has not previously been measured in randomized trials.
The superior pfs and efs in our second-line tki cohort might reflect selection bias at our institution.For instance, the number of non-smokers was statistically significantly higher in the tki group than in the docetaxel group (Table iii).Alternatively, follow-up imaging was not performed at a strict time interval (mean: 3.5 months), and a longer interval between treatment initiation and followup imaging could therefore have artificially prolonged pfs.The significance of such results in a retrospective study remains unclear, although it raises the interesting question of whether continuing tki beyond pfs could improve patient outcomes.Finally, differences in best response after firstline therapy and the variability of such clinical measures, even between large randomized trials (for example, os of 5.4 months in tailor and 9.0 months in delta), might account for the trend seen in our study favouring tki over docetaxel.
In the third-line setting, tkis do not seem to show a significant decrement in efficacy: our dcr of 32.1% suggests that, even in extensively pre-treated advanced nsclc, tkis are an interesting therapeutic option.
The side-effect profile of erlotinib has made it a quite attractive and tolerable option for patients who often have a high symptom burden and poor ecog performance status by the time they are eligible for second-line treatment.That situation is reflected in the present study: only 10 patients (7.8%) discontinued treatment because of severe side effects.Most patients stopped because of clinical disease progression.
In an effort to identify patients with EGFR wild-type nsclc who would most benefit from tki treatment, Taguchi and colleagues 17 developed a test that uses mass spectrometry analysis of serum to categorize candidates as likely to have good or poor survival on such treatment.Several retrospective studies have demonstrated that patients with a proteomic test classification of "good" experience a significantly better outcome than do those with a classification of "poor" when treated with egfr tkis [18][19][20] .A phase iii randomized trial by Gregorc et al. subsequently confirmed that patients with a proteomic test classification of "poor" experienced worse survival on erlotinib than on chemotherapy (hazard ratio: 1.72; 95% ci: 1.08 to 2.74; p = 0.022), while those with a classification of "good" experienced no significant difference in os with either treatment (adjusted hazard ratio: 1.06; 95% ci: 0.77 to 1.46; p = 0•714) 21 .Although mass spectrometry is promising, its integration into a resource-limited health care setting is less than certain, and the second-line treatment decision in EGFR wild-type nsclc remains largely clinical.
Findings in our study are limited by the relatively small size of the cohort and the retrospective nature of the analysis.Given the study's nature (real-life practice review), follow-up in the form of first imaging after the start of second-line treatment was performed at the clinician's discretion.Hence, some patients underwent imaging within weeks of treatment initiation, and others underwent imaging only months later when they demonstrated significant clinical progression.The median interval between pre-and post-treatment-initiation imaging was 3.5 months, and in 26 cases, patients did not undergo any imaging because they demonstrated early death or clinical progression and were on treatment for only a short time.We therefore chose to use efs to better describe the patients in whom treatment continued beyond pfs.The efs parameter has not been integrated into other larger randomized trials, and hence the significance of our efs findings, while intriguing, remains unclear.Nevertheless, the variability in the timing of imaging and the continuation of treatment both reflect real-life practice.Our data suggest that tkis are efficacious and well tolerated in the second-and third-line treatment of advanced nsclc.

CONCLUSIONS
Our study seems to show that second-line tki could be noninferior to docetaxel, suggesting a continuing role for targeted therapy in advanced nsclc.In the end, choice of second-and third-line therapy should be individualized based on numerous considerations, including symptom improvement, toxicity, patient comorbidities, convenience, and ease of administration.

FIGURE 1
FIGURE 1 Treatment trajectory in patients who received second-and third-line tyrosine kinase inhibitor (TKI).

FIGURE 2
FIGURE 2 Overall survival of patients who received tyrosine kinase inhibitor (TKI) or docetaxel in second-line treatment.Overall survival was defined as time elapsed from treatment initiation to death or study end.CI = confidence interval.
TKI IN SECOND-LINE TREATMENT OF EGFR WILD-TYPE NSCLC, Ma et al.

TABLE I
Results of selected randomized phase III trials examining the use of erlotinib after first-line treatment of non-small-cell lung cancer

TABLE III
a Second-line docetaxel compared with second-line TKI.BAC = bronchoalveolar carcinoma; ECOG PS = Eastern Cooperative Oncology Group performance status.

TABLE IV
Disease control rate with second-and third-line treatment using tyrosine kinase inhibitor (TKI) and docetaxel a Defined as clinical or radiologic evidence of progression, death from any cause, or discontinuation of TKI because of adverse events.

TABLE V
Tolerability of tyrosine kinase inhibitors (TKIS) in 145 patients