Patient and tumor characteristics
A total of 165 patients with newly diagnosed stage IV NSCLC were included, 92 in the send-out testing cohort and 73 in the in-house testing cohort (Fig. 1). Patient demographics were equally balanced among the send-out and in-house testing cohorts (Table 1). The median age of all patients was 72 years (range, 40-95 years) and 48% (n=80) were male. Most patients had lung adenocarcinoma (86%, n=142), while 7% (n=12) had squamous cell carcinoma, 0.6% (n=1) had large cell carcinoma, and 6% (n=10) had other/not otherwise specified pathology (Table 2). Molecular testing was most often performed on tissue biopsy or resection specimens (67%, n=111). There was a higher proportion of patients with testing on cytology specimens in the send-out cohort (42%, n=39) compared to the in-house testing cohort (21%, n=53; p=0.004).
Table 1 Baseline Demographics and Characteristics of Metastatic NSCLC Patients
|
Send-out Testing (n=92)
|
In-house Testing (n=73)
|
P-value
|
Age at diagnosis, years, median (IQR)
|
73 (66, 81)
|
70 (62, 78)
|
0.23
|
Sex, male
|
46 (50)
|
34 (46.6)
|
0.75
|
Ethnicity
|
|
|
0.58
|
Caucasian
|
40 (43.5)
|
25 (34.2)
|
|
Asian
|
23 (25)
|
24 (32.9)
|
|
Black
|
3 (3.3)
|
1 (1.4)
|
|
Other
|
4 (4.3)
|
2 (2.7)
|
|
Unknown
|
22 (23.9)
|
21 (28.8)
|
|
Smoking status
|
|
|
0.71
|
Current
|
20 (21.7)
|
19 (26)
|
|
Former
|
42 (45.7)
|
34 (46.6)
|
|
Never
|
30 (32.6)
|
20 (27.4)
|
|
ECOG Performance Statusa
|
|
|
0.11
|
0-1
|
48 (52.2)
|
44 (60.3)
|
|
≥2
|
37 (40.2)
|
19 (26)
|
|
Unknown
|
7 (7.6)
|
10 (13.7)
|
|
Age-adjusted CCI
|
|
|
0.21
|
≤3
|
44 (47.8)
|
27 (37)
|
|
>3
|
48 (52.2)
|
46 (63)
|
|
aECOG performance status was recorded at the time closest to treatment decision. Data are presented as n (%) unless indicated otherwise. Abbreviations: IQR = interquartile range; ECOG = Eastern Cooperative Oncology Group; CCI = Charlson Comorbidity Index. Statistical significance, two-sided p<0.05
Molecular biomarker testing patterns
The frequency of EGFR mutations detected in patients tested was 34% in both cohorts (send-out testing, n=32; in-house testing, n=25), and for ALK-positive cases was 7% (n=6) in the send-out testing cohort and 3% (n=2) in the in-house testing cohort (Table 2). The send-out cohort contained a higher proportion of cases with PD-L1 TPS <1% compared with the in-house testing cohort (35%, n=32 v 27%, n=20), and lower proportions of cases with TPS 1-49% (14%, n=13 v 26%, n=19) and ≥50% (39%, n=36 v 45%, n=33; p=0.01). Patients with send-out testing also had fewer PD-L1 tests ordered overall (88%, n=81 v 99%, n=72; p=0.01), as validation of PD-L1 testing on cytology cell block preparations was not completed until near the end of the send-out testing period.
Table 2 Pathological Tumor and Molecular Biomarker Testing Characteristics
|
Send-out Testing (n=92)
|
In-house Testing (n=73)
|
P-value
|
Histology
|
|
|
0.63
|
Adenocarcinoma
|
79 (85.9)
|
63 (86.3)
|
|
Squamous cell carcinoma
|
8 (8.7)
|
4 (5.5)
|
|
Large cell carcinoma
|
0 (0)
|
1 (1.4)
|
|
Other/NOS
|
5 (5.4)
|
5 (6.8)
|
|
Specimen type
|
|
|
0.004
|
Cytology
|
39 (42.4)
|
15 (20.6)
|
|
Tissue biopsy/resection
|
53 (57.6)
|
58 (79.4)
|
|
Sample inadequate for testing
|
11 (11.9)
|
5 (6.85)
|
0.27
|
EGFR
|
|
|
0.69
|
Positive
|
32 (34.8)
|
25 (34.2)
|
|
Negative
|
50 (54.3)
|
44 (60.3)
|
|
Indeterminate
|
3 (3.3)
|
1 (1.4)
|
|
Not ordered
|
7 (7.6)
|
3 (4.1)
|
|
ALK
|
|
|
0.34
|
Positive
|
6 (6.5)
|
2 (2.7)
|
|
Negative
|
79 (85.9)
|
67 (91.8)
|
|
Indeterminate
|
0 (0)
|
1 (1.4)
|
|
Not ordered
|
7 (7.6)
|
3 (4.1)
|
|
PD-L1
|
|
|
0.013
|
<1%
|
32 (34.8)
|
20 (27.4)
|
|
1-49%
|
13 (14.1)
|
19 (26)
|
|
≥50%
|
36 (39.1)
|
33 (45.2)
|
|
Not ordered
|
11 (12)
|
1 (1.4)
|
|
Data are presented as n (%). Abbreviations: NOS = not otherwise specified; EGFR = epidermal growth factor receptor; ALK = anaplastic lymphoma kinase; PD-L1 = programmed death-ligand 1. Statistical significance, two-sided p<0.05
Quality of testing
The percentage of unsuccessful biomarker tests due to insufficient tissue, suboptimal specimen quality (i.e. de-calcification, low cellularity), and sample processing issues was 12% (n=11) in the send-out testing cohort and 7% (n=5) in the in-house testing cohort (p=0.27; Table 2). Of those with initial inadequate samples, 7 patients in the send-out cohort and 3 patients in the in-house cohort underwent a second successful procedure to acquire additional tissue for biomarker analysis. Three patients in the send-out testing cohort underwent a second unsuccessful re-biopsy procedure, with 2 of these patients undergoing a third unsuccessful procedure. Of these patients, 1 patient proceeded with standard chemotherapy and 2 patients proceeded with best supportive care. The remaining 3 patients were too ill for repeated tissue biopsy and alternatively enrolled in a clinical trial that offered plasma-based molecular testing [23].
Availability of biomarker results
Of the NSCLC patients with in-house molecular biomarker testing, 88% (n=64) had test results available at their initial medical oncology consultation, compared to only 52% (n=48) of patients with send-out testing (p<0.0001; Fig. 2). All remaining patients with in-house testing had results for their last actionable biomarker available at the time of treatment decision (v 86%, n=38 with send-out testing; p=0.57), where an actionable biomarker is one that is associated with a directed systemic treatment (i.e. tyrosine kinase inhibitors for EGFR and ALK mutations, and immunotherapy with or without chemotherapy by PD-L1 expression levels). Six patients in the send-out testing cohort had treatment decisions made prior to reporting of complete actionable biomarker results. Two of these patients initiated treatment with chemotherapy, while the remaining 4 clinically deteriorated while awaiting their results and were ineligible for systemic therapy.
Workflow analysis
In this study, TTD is defined as the number of working days between the pathological diagnosis of NSCLC and the treatment decision for systemic therapy. Excluding patients whose treatment decisions were made before their biomarker results were reported, there was a significant improvement in TTD with the implementation of in-house biomarker testing from a median of 18 working days (IQR, 15-27 days) to 10 working days (IQR, 6-19 days; p<0.0001). To provide insight on the time-points that constituted the TTD, we performed a detailed workflow analysis of the pre-analytic, analytic, and post-analytic intervals (Table 3, Fig. 2).
The pre-analytic interval included the time from NSCLC diagnosis to biomarker testing request, and specimen transit time. Considering all biomarkers were reflexively ordered by the pathologist at the time of pathological diagnosis of NSCLC, the median time between sign-out of the diagnostic pathology report and opening of the biomarker case file in the Sunnybrook laboratory information system was 1 working day (IQR, 1 day) for both cohorts. Median specimen transit time, measured from biomarker testing request to the receipt and accessioning of the specimen in the laboratory for biomarker analysis, was faster for the in-house testing cohort than the send-out cohort (2 days [IQR, 1-2 days] v 3 days [IQR, 2-4 days]; p<0.0001).
The analytic interval, also referred to as laboratory turnaround time (TAT), is defined as the number of working days between the receipt of the sample in the laboratory for biomarker analysis and the release of the final biomarker report by the pathologist. With in-house testing, median laboratory TAT significantly decreased from 8 days (IQR, 7-11 days) to 3 days (IQR, 2-5 days; P<0.0001), with 96% of cases meeting the current CAP/IASLC/AMP guidelines of a ≤10-working day TAT versus 74% of cases with send-out testing (p=0.0002).
Implementation of in-house molecular biomarker testing did not impact the post-analytic interval, measured from the release of the final biomarker report to the treatment decision by the medical oncologist (Table 3).
All patients with in-house testing had their complete molecular biomarker report available at the time of treatment decision (Fig. 2). For the 6 patients with send-out testing whose treatment decisions were made prior to receiving their complete biomarker report, their workflow analysis is as follows: TTD, 6 days (IQR, 6-28 days); pathological diagnosis of NSCLC to testing request, 1 day (IQR, 1 day); specimen transit time, 5 days (IQR, 3-19 days); TAT, 10 days (8-13 days) with 67% of cases having a ≤10 working day TAT; receipt of biomarker report to treatment decision, -8 days (IQR, -9 to -2 days).
Table 3 Timeliness of Treatment Decisions for Metastatic NSCLC Patients
|
Send-out Testing (n=86)
|
In-house Testing (n=73)
|
P-value
|
Time to treatment decision, working days, median (IQR)
|
18 (15, 27)
|
10 (6, 19)
|
<0.0001
|
Diagnosis to testing request
|
1 (1, 1)
|
1 (1, 1)
|
0.85
|
Testing request to sample receipt by the lab
|
3 (2, 4)
|
2 (1, 2)
|
<0.0001
|
Sample receipt by the lab to biomarker report
|
8 (7, 11)
|
3 (2, 5)
|
<0.0001
|
≤10 days, n (%)
|
64 (74.4%)
|
70 (95.9%)
|
0.0012
|
>10 days, n (%)
|
22 (25.6%)
|
3 (4.1%)
|
|
Biomarker report to treatment decision
|
7 (5, 13)
|
7 (3, 12)
|
0.82
|
Note: Patients with treatment decisions made prior to release of their final biomarker report were excluded. Data are presented as median (IQR) unless indicated otherwise. Abbreviations: IQR = interquartile range. Statistical significance, two-sided p<0.05