PROGNOSTIC VALUE OF BIOMARKERS IN PATIENTS WITH NON-SMALL CELL LUNG CANCER HAVING STABLE DISEASE

Abstract

The present invention relates to an in vitro method for assessing the risk of non-small cell lung carcinoma (NSCLC) disease progression for a subject classified to have stable disease under an ongoing NSCLC treatment regime. The method involves determining the level of CYFRA 21-1 and/or the level of CA 125 in a sample obtained from the subject; and comparing (i) the determined level of CYFRA 21-1 to a CYFRA 21-1 cut-off level, (ii) the determined level of CA 125 to a CA 125 cut-off level, or (iii) a score taking into account the determined level of CYFRA 21-1 and/or the determined level of CA 125 to a cut-off score. The method of the invention further allows for assessing whether the subject responds to the ongoing treatment and/or whether the treatment regime should be maintained or modified. The invention also provides for corresponding uses, computer-implemented methods and computer program products.

Claims

1. An in vitro method for assessing the risk of non-small cell lung carcinoma (NSCLC) disease progression for a subject suffering from non-small cell lung carcinoma (NSCLC) and being under an ongoing NSCLC treatment regime, wherein the subject is classified to have a stable disease, and wherein said method comprises: a) determining a level of CYFRA 21-1 and/or a level of CA 125 in a sample obtained from the subject; and b) comparing (i) the determined level of CYFRA 21-1 to a CYFRA 21-1 cut-off level, (ii) the determined level of CA 125 to a CA 125 cut-off level, or (iii) a score taking into account the determined level of CYFRA 21-1 and/or the determined level of CA 125 to a cut-off score.

2. The method of claim 1, wherein in step (a) the level of CYFRA 21-1 is determined and the comparing in step (b) comprises (i) as defined in claim 1(b), wherein a determined level of CYFRA 21-1 lower than or equal to the CYFRA 21-1 cut-off level is indicative of a low risk of NSCLC disease progression; and/or wherein a determined level of CYFRA 21-1 higher than the CYFRA 21-1 cut-off level is indicative of a high risk of NSCLC disease progression.

3. The method of claim 2, wherein the subject suffers from NSCLC of subtype adenocarcinoma (ADC-NSCLC).

4. The method of claim 1, wherein in step (a) the level of CA 125 is determined and the comparing in step (b) comprises (ii) as defined in claim 1(b), wherein a determined level of CA 125 lower than the CA 125 cut-off level is indicative of a low risk of NSCLC disease progression; and/or wherein a determined level of CA 125 higher than the CA 125 cut-off level is indicative of a high risk of NSCLC disease progression.

5. The method of claim 4, wherein the subject suffers from NSCLC of subtype squamous cell carcinoma (SCC-NSCLC).

6. The method of claim 1, wherein in step (a) the levels of CYFRA 21-1 and/or CA 125 are determined and the comparing in step (b) comprises (iii) as defined in claim 1(b), and wherein a determined score lower than the cut-off score is indicative of a low risk of NSCLC disease progression; and/or wherein a determined score higher than the cut-off score is indicative of a high risk of NSCLC disease progression.

7. An in vitro method for assessing whether for a subject diagnosed with non-small cell lung carcinoma (NSCLC) an ongoing NSCLC treatment regime is to be maintained or modified and/or whether a subject diagnosed with non-small cell lung carcinoma (NSCLC) responds to an ongoing NSCLC treatment regime, wherein said subject is classified to have a stable disease, and wherein said method comprises: a) determining a level of CYFRA 21-1 and/or a level of CA 125 in a sample obtained from the subject; and b) comparing (i) the determined level of CYFRA 21-1 to a CYFRA 21-1 cut-off level, (ii) the determined level of CA 125 to a CA 125 cut-off level, or (iii) a score taking into account the determined level of CYFRA 21-1 and/or the determined level of CA 125 to a cut-off score.

8. (canceled)

9. The method of claim 1, wherein the method further comprises determining a level of CEA, and wherein the comparing in step (b) comprises (iii) as defined in claim 1(b), and wherein the score further takes into account the determined level of CEA.

10. The method of claim 1, wherein the method further comprises: obtaining an information whether the subject suffers from non-small cell lung carcinoma subtype squamous cell carcinoma (SCC-NSCLC) or adenocarcinoma (ADC-NSCLC); wherein the comparing in step (b) comprises (iii) as defined in claim 1(b), and wherein the score further takes into account the NSCLC subtype.

11. The method of claim 1, wherein the sample is a sample obtained 10 days to 150 days after the start date of the ongoing treatment regime.

12. The method of claim 1, wherein the sample is a blood sample.

13. (canceled)

14. (canceled)

15. (canceled)

16. (canceled)

17. A kit comprising: a reagent or a set of reagents for detecting a level of CYFRA 21-1 and/or a reagent or a set of reagents for detecting a level of CA 125, and optionally a reagent or set of reagents for detecting a level of CEA in a sample obtained from a subject, wherein the kit is a kit for assessing: (i) the risk of NSCLC disease progression under an ongoing NSCLC treatment regime for a subject diagnosed with NSCLC; (ii) whether a subject diagnosed with NSCLC responds to an ongoing NSCLC treatment regime; and/or (iii) whether for a subject diagnosed with NSCLC an ongoing NSCLC treatment regime is to be maintained or modified; wherein said subject of (i), (ii), and (iii) is classified to have a stable disease.

18. The method of claim 7, wherein the method further comprises determining a level of CEA, and wherein the comparing in step (b) comprises (iii) as defined in claim 7(b), and wherein the score further takes into account the determined level of CEA.

19. The method of claim 7, wherein the method further comprises: obtaining an information whether the subject suffers from non-small cell lung carcinoma subtype squamous cell carcinoma (SCC-NSCLC) or adenocarcinoma (ADC-NSCLC); wherein the comparing in step (b) comprises (iii) as defined in claim 7(b), and wherein the score further takes into account the NSCLC subtype.

20. The method of claim 10, wherein the score further takes into account the NSCLC subtype by using an interaction term between the information about the NSCLC subtype and the level of CYFRA 21 and/or the level of CA 125.

21. The method of claim 11, wherein the sample is a sample obtained 20 days to 120 days after the start date of the ongoing treatment regime.

22. The method of claim 11, wherein the sample is a sample obtained 25 days to 108 days after the start date of the ongoing treatment regime.

23. The method of claim 12, wherein the sample is a blood sample selected from the group consisting of whole blood, serum, and plasma.

24. The method of claim 1, wherein said subject is a subject classified to have a stable disease based on imaging data.

25. The method of claim 1, wherein the ongoing NSCLC treatment regime is selected from chemotherapy, targeted therapy, and immunotherapy, and combinations thereof.

Description

DESCRIPTION OF THE FIGURES

[0511] The following Figures are provided to aid the understanding of the present invention, the true scope of which is set forth in the appended claims. It is understood that modifications can be made in the procedures set forth without departing from the spirit of the invention.

[0512] FIG. 1: Progression-free survival (A) and overall survival (B) in patients with partial response or stable disease at the first CT scan after the second cycle stratified by CT scan results.

[0513] FIG. 2: Progression-free survival in patients with stable disease at the first CT scan after the second cycle in those with adenocarcinoma stratified by CYFRA 21-1 (A), squamous cell carcinoma stratified by CA 125 (B), adenocarcinoma or squamous cell carcinoma stratified by CYFRA 21-1 and CA 125 (C), and adenocarcinoma or squamous cell carcinoma stratified by CYFRA 21-1, CA 125 and CEA (D), above or below median. Biomarker combinations by score (referred to as Pred in the Figures) in (C) and (D) take into account interaction terms between each biomarker and the histology. No baseline correction for biomarker levels was used.

[0514] FIG. 3: Progression-free survival (A) and overall survival (B) in patients with stable disease at the first CT scan after the second cycle in those with adenocarcinoma or squamous cell carcinoma stratified into high- and low-risk groups by CYFRA 21-1, CA 125 and CEA, above or below median. The model for the biomarker combinations takes into account interaction terms between each biomarker and the histology. No baseline correction for biomarker levels was used.

[0515] FIG. 4: Progression-free survival (A) and overall survival (B) in patients with stable disease at the first CT scan after the second cycle in those with adenocarcinoma or squamous cell carcinoma stratified into high- and low-risk groups by CYFRA 21-1, CA 125 and CEA, above or below optimized cut-off. The model for the biomarker combinations takes into account interaction terms between each biomarker and the histology. No baseline correction for biomarker levels was used.

EXAMPLES

[0516] The following examples are provided to aid the understanding of the present invention, the true scope of which is set forth in the appended claims. IT is understood that modifications can be made in the procedures set forth without departing from the spirit of the invention.

Example 1: Assessment of the Prognostic Value of a Serum Biased Biomarkers in Patients with NSCLC

[0517] To study whether blood based tumor biomarkers have predictive value in monitoring treatment success and disease progression prognosis as such or in addition to imaging based analyses, a clinical study was conducted involving parallel state-of-the art imaging based tumor staging/treatment monitoring and parallel measurement of seven pre-selected biomarker candidates (CEA, ProGRP, NSE, CYFRA 21-1, SCC, CA 15-3, and CA 125).

[0518] The cohort of this clinical study includes patients ≤18 years of age with previously untreated Stage III or IV NSCLC (adenocarcinoma or squamous cell carcinoma (SCC) histology) and Eastern Cooperative Oncology Group (ECOG) performance score 0-2.

[0519] Of the 387 NSCLC patients, 265 patients received first-line treatment, did not have progressive disease before treatment Cycle 2 (Cycle 2) and had computed tomography (CT) data available 25 days to 108 days (median 44 days) after treatment start.

[0520] At the first CT after Cycle 2, 230 patients did not have progressive disease (100 patients had stable disease [SD] and 130 patients had a partial response [PR]). For 228 of these patients (100 patients with SD and 128 patients with PR) complete biomarker results, i.e. levels of CEA, ProGRP, NSE, CYFRA 21-1, SCC, CA 15-3, and CA 125 were available.

[0521] Exclusion criteria were: inability to obtain blood samples, history of secondary malignancy, severe co-morbidities, any type of tumor-directed pre-treatment (some patients received radiotherapy defined first-line treatment), pregnancy or breastfeeding.

[0522] Patients received as first line therapy chemotherapy (75.1% of all patients), tyrosine kinase inhibitors and/or immune checkpoint inhibitors at the discretion of the treating physician.

[0523] Patients with NSCLC (adenocarcinoma or SCC) and available CT scan data after the second treatment cycle of the first line therapy were used for the biomarker analysis (see below) Patients who had documented disease progression before the first CT scan were excluded from the analysis with the biomarker data.

[0524] Patient demographics and treatment schemes of the analyses population are summarized in Tables 1a, 1b and 1c, below.

TABLE-US-00003 TABLE 1a Patient demographics and treatment schemes (analyses population); All values n (%), unless otherwise stated. Adeno, adenocarcinoma; CT, computed tomography; ECOG, Eastern Cooperative Oncology Group, ICI, immune checkpoint inhibitor; PR, partial response; SCC, squamous cell carcinoma; SD, stable disease; TKI, tyrosine kinase inhibitor; UICC, Union for International Cancer Control. Patients with SD at first CT after Cycle 2 Adeno SCC All (n = 76) (n = 24) (N = 100) Mean age (standard 67.3 (8.9) 67.6 (8.2) 67.3 (8.7) deviation) Male sex 47 (61.8) 22 (91.7) 69 (69.0) Smoking status Smoker 26 (34.2) 8 (33.3) 34 (34.0) Ex-smoker 42 (55.3) 16 (66.7) 58 (58.0) Never smoked 8 (10.5) — 8 (8.0) UICC stage III 16 (21.1) 11 (45.8) 27 (27.0) IV 60 (79.0) 13 (54.2) 73 (73.0) ECOG performance status 0 37 (48.7) 13 (54.2) 50 (50.0) 1 38 (50.0) 11 (45.8) 49 (49.0) 2 1 (1.3) — 1 (1.0) Treatment until first CT after Cycle 2 Platinum-based 48 (63.2) 23 (95.8) 71 (71.0) combination Platinum-free 6 (7.9) — 6 (6.0) TKI ± radiotherapy 8 (10.5) — 8 (8.0) ICI alone or in 11 (14.5) 1 (4.2) 12 (12.0) combination No treatment 3 (4.0) — 3 (3.0)

TABLE-US-00004 TABLE 1b Patient demographics and treatment schemes (analyses population); All values n (%), unless otherwise stated. Adeno, adenocarcinoma; CT, computed tomography; ECOG, Eastern Cooperative Oncology Group, ICI, immune checkpoint inhibitor; PR, partial response; SCC, squamous cell carcinoma; SD, stable disease; TKI tyrosine kinase inhibitor; UICC, Union for International Cancer Control. Parents with PD at first CT after Cycle 2 Adeno SCC All (n = 27) (n = 8) (N = 35) Mean age (standard 62.3 (9.0) 66.0 (5.3) 63.1 (8.4) deviation) Male sex 18 (66.7) 4 (50.0) 22 (62.9) Smoking states Smoker 15 (55.6) 4 (50.0) 19 (54.3) Ex-smoker 10 (37.0) 3 (37.5) 13 (37.1) Never smoked 2 (7.4) 1 (12.5) 3 (8.6) UICC stage III 1 (3.7) 2 (25.0) 3 (8.6) IV 26 (96.3) 6 (75.0) 32 (91.4) ECOG performance status 0 14 (51.9) 3 (37.5) 17 (48.6) 1 13 (48.2) 5 (62.5) 18 (51.4) 2 — — — Treatment until first CT after Cycle 2 Platinum-based 20 (74.1) 8 (100) 28 (80.0) combination Platinum-free — — — TKI ± radiotherapy 1 (3.7) — 1 (2.9) ICI alone or in 5 (18.5) — 5 (14.3) combination No treatment 1 (3.7) — 1 (2.9)

TABLE-US-00005 TABLE 1c Patient demographics and treatment schemes (analyses population); All values n (%), unless otherwise stated. Adeno, adenocarcinoma; CT, computed tomography; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; PR, partial response: SCC, squamous cell carcinoma; SD, stable disease; TKI, tyrosine kinase inhibitor, UICC, Union for International Cancer Control. Patients with PR at first CT after Cycle 2 Adeno SCC All (n = 90) (n = 38) (N = 128) Mean age (standard 65.3 (10.5) 67.7 (7.8) 66.0 (9.8) deviation) Male sex 43 (47.8) 31 (81.6) 74 (57.8) Smoking status Smoker 28 (31.1) 16 (42.1) 44 (34.4) Ex-smoker 45 (50.0) 21 (55.3) 66 (51.6) Never smoked 17 (18.9) 1 (2.6) 18 (14.1) UICC stage III 18 (20.0) 23 (60.5) 41 (32.0) IV 72 (80.0) 15 (39.5) 87 (68.0) ECOG performance status 0 43 (47.8) 21 (55.3) 64 (50.0) 1 47 (52.2) 17 (44.7) 64 (50.0) 2 — — — Treatment until first CT after Cycle 2 Platinum-based 53 (58.9) 36 (94.7) 89 (69.5) combination Platinum-free 3 (3.3) 1 (2.6) 4 (3.1) TKI ± radiotherapy 20 (22.2) 1 (2.6) 21 (16.4) ICI alone or in 9 (10.0) — 9 (7.0) combination No treatment 5 (5.6) — 5 (3.9)

Study Objectives

[0525] The main objective was to assess the predictive value of serum biomarkers in monitoring treatment success and the prognosis for disease progression. Specifically, it was assessed whether patients with stable disease (SD) based on CT can be further differentiated into risk groups (e.g. responders with good prognosis) based on serum biomarker levels. This analysis was conducted after the first two cycles of treatment. The measures used for the assessment were overall survival (OS) and progression free survival (PFS), each of them individually analyzed.

Treatment Efficacy Assessment by State of the Art CT Imaging

[0526] The first CT scan after Cycle 2 of first-line treatment was set as the analysis time point for all patients (maximum time between the second cycle treatment and the first CT after Cycle 2 was 60 days). The tumor response on CT was defined by the response evaluation criteria in solid tumors (RECIST 1.1: see Eisenhauer et al., Eur J Cancer 2009, 45, 228-247. doi:10.1016/j.ejca.2008.10.026). Partial Response (PR) was assigned to >=30% decrease in size. Progressive Disease (PD) was assigned to >=20% increase in size. Stable Disease was assigned for such cases were neither sufficient decrease in size to qualify as PR nor sufficient growth to qualify as a PD was found (i.e. <30% decrease in size to <20% increase in size).

[0527] Progression-free survival (PFS) time and overall survival (OS) time were calculated from treatment start date. PFS was defined as follows: [0528] Progression=Patient has at least one progression (i.e. PD according to RECIST) observed or died; [0529] No progression=Patient has no progression and was alive at study completion; [0530] PFS time for Patients with Progression=Date of progression (detected)−Therapy start date+1 (in days); [0531] PFS time for patients that died during study=Death date−Therapy start date+1 (in days); [0532] PFS time for patients without progression event=Last contact date−Therapy start date+1 (in days).

[0533] OS time was defined as follows: [0534] OS time for patients that died during the study=Death date−Therapy start date+1 (in days); [0535] OS time for patients alive after last contact=Last contact date−Therapy start date+1 (in days).

Sample Collection and Biomarker Evaluation

[0536] Venous blood samples were collected from each patient at baseline (i.e. before treatment start) and each routine visit (usually at each treatment cycle (approx. 21 days each)), if possible. The samples for the analysis at the first CT after the 2.sup.nd treatment cycle were taken in a range of 10 days before to the CT to 29 days after the CT. Calculated from the day of the treatment start the samples were taken 25 to 108 days after that date. Serum samples were stored as 500 μl aliquots below −70° C.

[0537] In these serum samples biomarker levels were measured at a later timepoint. Specifically, electrochemiluminescence immunoassays (ECLIA) for in vitro diagnostics protein biomarkers CEA, ProGRP, NSE, CYFRA 21-1, SCC, CA 15-3, and CA 125 were conducted on Cobas® systems from Roche Diagnostics Centralized and Point of Care Solutions (CPS) according to the manufacturer's guidelines.

Statistical Analysis

[0538] Patient demographics and disease characteristics were summarized using descriptive statistics. Risk prediction for progression (PFS or OS) was compared between CT response and biomarker values using Cox regression models and Kaplan-Meier curves.

[0539] Prognostic models were based on response at first CT after Cycle 2 (PR versus SD), biomarker values at first CT after Cycle 2 and biomarker change between baseline and first CT (baseline correction) after Cycle 2, respectively. Biomarker values were log 2 transformed. For univariate models (including only one biomarker) a cutoff based on the value of the biomarker is used to split the patients in a low and a high risk group, e.g. the median of the biomarker is used. For illustration the cut-offs of the univariate models were backtransformed to the original biomarker scale for the Kaplan-Meier curves. For models including more than one biomarker a score was built by weighted linear combination based on the linear predictor of the Cox regression model. To split the patients in a low and a high risk group e.g. the median of this score was used equally to the univariate biomarker.

[0540] Further, in alternative, for models including more than one biomarker a binary-score was built based on the combination of the risk groups of the univariate models of each included biomarker. In brief, similar to the univariate models (including one biomarker) a cut-off based on the on the value of each individual biomarker was defined (e.g. the median of the biomarker levels). If both biomarker levels were above the respective biomarker cut-offs forming pan of the binary score cut-off, the patients were categorized to the high risk group. If a single biomarker was below the respective cut-off; i.e. one of the levels of the binary score was below the corresponding cut-off value of the binary score cut-off, the patients were categorized to the low risk group.

[0541] Prognostic models were evaluated using hazard ratios (HRs) and C-Indexes. The C-Index is a non-parametric estimator of the proportion of all patient pairs for which model prediction and observed outcome are concordant and is therefore a global evaluation criteria for the Cox regression model. A C-Index of 1 corresponds to the best model prediction and a C-Index of 0.5 represents a random prediction. A hazard ratio describes the risk between patient groups, e.g. a hazard ratio of 2 means that the risk of a patient in the high risk group is two times higher compared to a patients in the low risk group.

[0542] In biomarker-based risk prediction models, patients of the SD group were separated into two risk groups based using either the median biomarker/score value as cut-off or an optimized cut-off. The optimized cut-off was determined in that all quantiles from 0.2 to 0.8 by steps of 0.05 of the biomarker/score value were tested for the performance to split the patients into two groups and the hazard ratio and log-rank p-value for each split was calculated. The quantile with the lowest log-rank p-value was chosen as the optimized cut-off.

[0543] Cox proportional hazard models were based on a single biomarker or combinations of two or three biomarkers. For combined population of adenocarcinoma and SCC (histology information), the models also include an interaction term (Vatcheva, K. P., et al. Epidemiology (Sunnyvale, Calif.) 6.1, 2015) between said histology information and each biomarker to account for the association between histology and a respective biomarker. An example for a model formula including three biomarkers and the interaction term between the biomarker and the histology of the Cox-regression is:

h(t)=h.sub.0(t)*exp(β.sub.1*log 2(BM.sub.1)+β.sub.2*log 2(BM.sub.1)×SCC+β.sub.3*log 2(BM.sub.2)+β.sub.4*log 2(BM.sub.2)*SCC+β.sub.5*log 2(BM.sub.3)+β.sub.6*log 2(BM.sub.3)*SCC)

[0544] With representing the survival time, h(t) representing the hazard function, the coefficients β.sub.1, . . . , β.sub.6 measuring the impact of the covariates, BM.sub.1, . . . , BM.sub.3 representing the biomarkers and SCC representing the histology squamous cell carcinoma. For patients without SCC the formula reduces as SCC is set to 0:

h(t)=h.sub.0(t)*exp(β.sub.1*log 2(BM.sub.1)+β.sub.3*log 2(BM.sub.2)+β.sub.5*log 2(BM.sub.3))

[0545] For patients with SCC the whole formula is taken into account as SCC is set to 1.

[0546] Similar to the model including multiple biomarkers a score was built based on the risk prediction of the Cox regression model.

Results of Study

[0547] a) Prognostic Value of State-of-the Art First CT Scan after Second Treatment Cycle

[0548] First, the prognostic value of the first CT scan after the second treatment cycle in predicting progression-free survival (PFS) or overall survival (OS) was assessed. Monitoring by CT is the current state of the art in assessing NSCLC progression and treatment response.

[0549] Patients identified as having PR or SD at the first CT scan had similar risk of progression (see FIG. 1A), suggesting that the first CT scan had poor prognostic performance for PFS in patients deriving clinical benefit from the ongoing treatment regime, i.e. patients of the PR and SD groups.

[0550] The specific results regarding PFS are as follows: [0551] Adenocarcinoma or SCC: Hazard ratio (HR)=1.326 (p-value=0.055), C-index=0.579. [0552] Adenocarcinoma: HR=1.482 (p-value=0.025), C-Index: 0.616. [0553] SCC: HR=0.992 (p-value=978), C-Index: 0.507.

[0554] The first CT scan also showed poor prognostic performance for OS, with rates of OS largely similar in patients with PR and SD (FIG. 1B).

[0555] The specific results regarding OS are: [0556] Adenocarcinoma or SCC: HR=1.517 (p-value=0.012), C-index=0.608. [0557] Adenocarcinoma: HR=1.784 (p-value=0.005), C-Index 0.656 SCC: 0.924 (p-value=0.786), C-Index 0.527

b) Prognostic Value of Cancer Biomarkers Using Non-Baseline Corrected Single Measurement Values

[0558] The prognostic value of individual cancer biomarkers (CA 125, CA 15-3, CEA, CYFRA 21-1, NSE. SCC and proGRP), or combinations of cancer biomarkers, in predicting PFS or OS were assessed in patients with SD using single protein biomarker levels in serum without baseline correction (i.e. absolute biomarker level without taking the biomarker concentration before treatment into account).

Analyses Based on Median Biomarker Cut-Off

[0559] In a first set of analyses the biomarker data (non-baseline corrected) were analyzed by splitting SD patients into two groups at the median value for each biomarker; i.e. using the median of the absolute biomarker value or the median of the combined biomarker value (score) as cut-off. The performance of single biomarkers (univariate analyses) and of several biomarkers combined to a score (multivariate analyses) was assessed.

[0560] The results for the read-out progression-free survival are summarized in Table 2, below:

TABLE-US-00006 TABLE 2 Results of univariate and combination analyses of the prognostic values of CYFRA 21-1, CA 125, CEA, CA15-3, NSE, SCC and ProGRP for progression- free survival in patients with stable disease at the first CT scan after the second cycle. Higher C-index indicates better performance of the Cox regression model, higher HR indicates better performance in separating the SD population in high and low risk populations. Adeno + SCC with Adeno + SCC interaction term Adeno (n = 76) SCC (n = 24) (n = 100) (n = 100) C- HR C- HR C- HR C- HR index median.sup.a index median.sup.a Index median.sup.a index median.sup.a Univariate analysis CYFRA 21-1 0.689 2.197 0.606 2.878 0.667 2.293 0.671 2.405 CA 125 0.626 1.897 0.680 4.087 0.619 1.707 0.632 2.492 CEA 0.597 1.547 0.509 1.213 0.561 1.270 0.585 1.503 CA15-3 0.557 1.141 0.563 1.087 0.554 1.162 0.559 1.189 NSE 0.535 1.147 0.506 1.043 0.528 1.019 0.537 1.141 SCC 0.520 1.219 0.610 1.306 0.538 1.229 0.545 1.103 proGRP 0.507 0.834 0.498 0.821 0.507 0.826 0.527 1.498 Combination analysis CYFRA 21-1 + 0.689 2.633 0.688 5.354 0.666 2.758 0.678 2.910 CA 125 CYFRA 21-1 + 0.708 2.790 0.695 5.354 0.674 2.412 0.702 2.372 CA 125 + CEA .sup.aPatients stratified according to biomarker value above or below median. Adeno, adenocarcinoma; HR, hazard ratio; SCC, squamous cell carcinoma.

[0561] Of the single biomarkers tested, CYFRA 21-1 had the highest prognostic value for adenocarcinoma, in patients with SD (see Table 2 and FIG. 2A) as indicated by C-index and HR. A combination of CYFRA 21-1 and CA 125 showed an improved performance in separating the groups of high and low risk for disease progression, as indicated by a higher HR. The best performance was achieved by a combination analysis (score) involving CYFRA 21-1, CA 1-25 and CEA.

[0562] For SCC patients, CA 125 had the highest prognostic value in patients with SD as single biomarker (see Table 2 and FIG. 2B) as indicated by C-index and HR. A combination of CYFRA 21-1 and CA 125 showed an improved performance in C-index and HR. The best performance was achieved by a combination analysis (score) involving CYFRA 21-1, CA 125 and CFA.

[0563] For the combined population of adenocarcinoma and SCC, a combination of CYFRA 21-1 and CA 125 (using an interaction term) had a greater prognostic value than either biomarker alone in patients with SD according to RECIST (see Table 2 and FIG. 2C). Building a score taking into account CYFRA 21-1. CA 125, CEA and in addition an interaction term between the biomarkers and the histology (SCC or adenocarcinoma),) the prognostic performance of the combination for adenocarcinoma and SCC in the patient cohort with SD NSCLC could be further increased (see Table 2):

[0564] Notably, Patients with SD, and a lower risk of progression or death based on the individual biomarkers CYFRA 21-1 or CA 125 or the combination model including CYFRA 21-1, CA 125 and CEA (and optionally an interaction term), had a similar probability for progression compared with patients with PR as indicated by hazard ratios (HR) (FIG. 3, Table 3). To calculate those hazard ratios, a variable was created which is “PR” for patients with PR, “low risk” for SD patients in the low risk group and “high risk” for SD patients in the high risk group. A further Cox model including this covariate was fitted, this results in the H R between each combination of the three groups (“low risk” vs. “high risk”, “low risk” vs. PR and “high risk” vs. PR).

TABLE-US-00007 TABLE 3 Results of univariate and combination analyses of the prognostic values of CYFRA 21-1, CA 125 and CEA for progression-free survival in patients with stable disease at the first CT scan after the second cycle. A HR near to 1 indicates a similar risk for progression between a patient with SD in the low risk group and a patient with PR. Adeno + SCC with Adeno + interaction Adeno SCC SCC term (n = 76 (n = 24) (n = 100) (n = 100) HR low risk HR low risk HR low risk HR low risk vs. PR vs. PR vs. PR vs. PR Univariate analysis CYFRA 21-1 0.972 1.482 1.096 1.130 CA 125 0.895 1.621 0.953 1.081 CEA 0.848 1.138 0.852 0.919 CA15-3 0.727 1.027 0.809 0.828 NSE 0.725 1.045 0.761 0.807 SCC 0.740 1.106 0.827 0.783 proGRP 0.609 0.895 0.681 0.920 Combination analysis CYFRA 21-1 + 1.059 1.642 1.155 1.189 CA 125 CYFRA 21-1 + 1.056 1.642 1.102 1.107 CA 125 + CEA .sup.aPatients stratified according to biomarker value above or below median. Adeno, adenocarcinoma; HR, hazard ratio; SCC, squamous cell carcinoma.

[0565] For PFS, the HR for high versus low risk patients (based on the score taking into account CYFRA 21-1 CA 125, CEA and the histology interaction term) amounts to 2,372 (p-value<0.001), indicating that the risk for disease progression is significantly higher for high risk patients according to the biomarker model (see Table 2). The HR for PR versus SD in the low risk based on this score amounts to 1.107 (p-value 0.594), indicating that the SD patients categorized as low risk for disease progression by the biomarker model have a comparable risk for disease progression than PR patients according to imaging (FIG. 3A and Table 3). Similar analyses made for the first read out PFS have also been made for the second read-out OS.

[0566] The data on OS (see Table 4) confirmed the findings based on the PFS readout.

[0567] Again, a robust prognostic separation of patients at high risk for disease progression/death and low risk for disease progression/death could be achieved based on the same biomarkers and scores.

[0568] For OS, the HR for high versus low risk patients (based on the score taking into account CYFRA 21-1 CA 125, CEA and the histology interaction term) amounts to 2.091 (p-value 0.002), again confirming that also for this read out the separation between high and low risk works well using the biomarker model (see Table 4). The HR for PR versus low risk based on this score amounts to 0.960 (p-value 0.853), confirming that the low risk group identified by the biomarker model has a comparable risk to the PR patient group (see Table 5). The HR for PD versus high risk amounts to 1.760 (FIG. 3B.

TABLE-US-00008 TABLE 4 Results of univariate and combination analyses of the prognostic values of CYFRA 21-1, CA 125 and CEA for overall survival in patients with stable disease at the first CT scan after the second cycle. Higher C-index indicates better performance of the Cox regression model, higher HR indicates better performance in separating the SD population in high and low risk populations. Adeno + SCC with Adeno + SCC interaction term Adeno (n = 76) SCC (n = 24) (n = 100) (n = 100) C- HR C- HR C- HR C- HR index median.sup.a index median.sup.a Index median.sup.a index median.sup.a Univariate analysis CYFRA 21-1 0.713 2.771 0.609 1.305 0.694 2.363 0.692 2.357 CA 125 0.659 2.561 0.617 3.827 0.640 2.040 0.649 2.573 CEA 0.559 1.119 0.512 0.877 0.545 1.117 0.545 1.394 CA15-3 0.583 1.424 0.672 1.400 0.599 1.575 0.599 1.549 NSE 0.544 1.373 0.528 1.206 0.542 1.130 0.525 1.246 SCC 0.528 1.248 0.540 1.123 0.532 1.366 0.541 1.303 proGRP 0.517 1.175 0.533 0.878 0.522 1.123 0.516 1.081 Combination analysis CYFRA 21-1 + 0.729 3.110 0.617 1.673 0.705 2.541 0.703 2.540 CA 125 CYFRA 21-1 + 0.715 2.740 0.665 1.527 0.703 2.354 0.698 2.091 CA 125 + CEA .sup.aPatients stratified according to biomarker value above or below median. Adeno, adenocarcinoma; HR, hazard ratio; SCC, squamous cell carcinoma.

TABLE-US-00009 TABLE 5 Results of univariate and combination analyses of the prognostic values of CYFRA 21-1, CA 125 and CEA for overall survival in patients with stable disease at the first CT scan after the second cycle. A HR near to 1 indicates a similar risk for progression between a patient with SD in the low risk group and a patient with PR. Adeno + SCC with Adeno + Interaction Adeno SCC SCC term (n = 76) (n = 24) (n = 100) (n = 100) HR low risk HR low risk HR low risk HR low risk vs. PR vs. PR vs. PR vs. PR Univariate analysis CYFRA 21-1 0.906 1.211 0.989 1.015 CA 125 0.900 1.842 0.932 1.031 CEA 0.610 1.014 0.706 0.787 CA15-3 0.675 1.218 0.811 0.815 NSE 0.677 1.263 0.715 0.747 SCC 0.639 1.199 0.785 0.765 proGRP 0.623 1.038 0.706 0.701 Combination analysis CYFRA 21-1 + 0.978 1.272 1.032 1.045 CA 125 CYFRA 21-1 + 0.916 1.259 0.994 0.960 CA 125 + CEA .sup.aPatients stratified according to biomarker value above or below median. Adeno, adenocarcinoma; HR, hazard ratio; SCC, squamous cell carcinoma.

[0569] In sum, the above results demonstrate that CYFRA 21-1 and CA 125 absolute levels as such, a score combining the absolute levels of the two biomarkers and in particular a score taking into account CYFRA 21-1, CA 125 and CEA absolute levels can provide additional guidance to the first CT scan after the second treatment cycle in patients with indeterminate CT response, by differentiating those with SD into high and low risk groups. This was validated by using PFS and OS. Further, we have demonstrated that the biomarker score can be further improved by taking into account an interaction term based on tumor histology (SCC or adenocarcinoma) and optimizing the cut-off.

Analyses Based Optimized Biomarker Cut-Off

[0570] The analysis using the model/score based on CYFRA 21-1, CA 125 and CEA and using the interaction term was repeated with an optimized cut-off Specifically, the cut-off was optimized to obtain the maximum difference in risk between the biomarker groups by keeping 20% of the patients in each of the two risk groups (high and low risk group).

[0571] With the optimized cut-off, patients with high-risk SD according to the biomarker model had a similar survival probability to those with PD for OS, and patients with low-risk SD according to the biomarker model had a similar survival probability to those with PR for PFS and OS. (FIG. 4). Cut-off optimization was able to further increase the IR for the high vs low risk groups underlining the improved prognostic differentiation that can be made by using the biomarker score.

[0572] For PFS, the HR for high versus low risk of the model including CYFRA 21-1, CA125, CEA and the interaction term amounted to 3.241. The HR for PR versus low risk amounted to 1.023 (FIG. 4A).

[0573] For OS, HR for high versus low risk of the model including CYFRA 21-1. CA125, CEA and the interaction term amounted to 4.206. The HR for PR versus low risk amounted to 0.924. The HR for PD versus high risk amounted to 0.940 (FIG. 4B).

Analyses Based on Combination of Individual Biomarker Splits (Binary Score)

[0574] The analysis using the model/score based on CYFRA21-1 and CA125 using the optimized cutoff was repeated by applying the optimized cutoff (as described above) to the single biomarkers, separating all patients into a low-risk SD and high-risk SD group based on the single marker and then combining the splits of the two biomarkers. In a nutshell, a binary score was built comprising a cut-off for CYFRA21-1 and CA125, respectively, and the CYFRA21-1 level and CA125 level were compared to the respective cut-offs, respectively. If a biomarker level was higher than the cut-off, the respective part of the binary score was defined as high. If a biomarker level was below or at the cut-off, the respective part of the binary score was defined as low. The resulting groups of the binary score based on the combination of the individual splits were then further summarized in a group with high risk (CYFRA21-1 above cutoff and CA125 above cutoff) versus low risk (CYFRA21-1 below cutoff and CA125 below cutoff, CYFRA21-1 below cutoff and CA125 above cutoff, CYFRA21-1 above cutoff and CA125 below cutoff.

[0575] With the optimized cut-off, patients with high-risk SD according to the biomarker model had a similar survival probability to those with PD for OS, and patients with low-risk SD according to the biomarker model had a similar survival probability to those with PR for PFS and OS.

[0576] For PFS based on all patients, the HR for high versus low risk of the model including CYFRA 21-1 and CA125 term amounted to 3,430. The HR for PR versus low risk amounted to 0.856.

[0577] For OS based on all patients, the HR for high versus low risk of the model including CYFRA 21-1 and CA125 amounted to 3.341. The HR for PR versus low risk amounted to 0.796. The HR for PD versus high risk amounted to 0.913. Accordingly, the binary score delivers similar results as a score using linear combination.

c) Prognostic Value of Cancer Biomarkers Using Baseline Correction of Biomarker Levels with the Biomarker Levels Before Treatment.

[0578] The results obtained by analyses using baseline corrected biomarker levels (i.e. ratio of the biomarker levels at first CT after second treatment cycle and initial biomarker concentration before treatment) for progression free survival are summarized in Table 6 below. The baseline corrected biomarker levels are log 2 transformed.

TABLE-US-00010 TABLE 6 Results of univariate and combination analyses of the prognostic values of the ratio of CYFRA 21-1, CA 125 and CEA to BL for progression free survival in patients with stable disease at the first CT scan after the second cycle. Higher C-index indicates better performance of the Cox regression model, higher HR indicates better performance in separating the SD population in high and low risk populations. Adeno + SCC with Adeno + SCC interaction term Adeno (n = 76) SCC (n = 24) (n = 100) (n = 100) C- HR C- HR C- HR C- HR index median.sup.a index median.sup.a Index median.sup.a index median.sup.a Univariate analysis CYFRA 21-1 0.461 0.830 0.454 1.617 0.522 0.785 0.482 1.057 CA 125 0.513 0.907 0.610 2.749 0.535 1.115 0.538 0.979 CEA 0.596 2.086 0.487 1.076 0.576 1.786 0.581 1.624 Combination analysis CYFRA 21-1 + 0.510 0.777 0.613 2.749 0.553 1.092 0.538 0.945 CA 125 CYFRA 21-1 + 0.602 2.102 0.617 2.816 0.576 1.486 0.597 1.907 CA 125 + CEA .sup.aPatients stratified according to biomarker value above or below median. Adeno, adenocarcinoma;; HR, hazard ratio; SCC, squamous cell carcinoma.

[0579] The results surprisingly show that the data without baseline correction relying on the absolute level of the biomarker (see table 2) show much better results in differentiating the SD patient group into high and low risk patients for PFS and OS than using the biomarker level changes (see table 6). It is a true advantage and a surprising finding that the risk assessment and, thus, the decision whether the patient benefits from the treatment can be based on a single biomarker measurement and does not require a baseline measurement. Even more surprising is that the performance of single biomarker measurements and scores derived therefrom is by far superior compared to the same biomarkers yet based on a change in the biomarker(s) level(s) compared to baseline (before treatment).

CONCLUSIONS

[0580] CT is currently the method of choice for assessing response to therapy in lung cancer; however, in this study patients with PR at the first CT scan after the second treatment cycle had a similar risk of progression and survival to patients with SD. Furthermore, CT scan and the categorical RECIST criteria do not permit risk discrimination within patients with SD.

[0581] Of 7 biomarkers tested, the optimal prognostic biomarker differed depending on histology of the tumor; CYFRA 21-1 had the highest prognostic value for adenocarcinoma NSCLC, and CA 125 for SCC NSCLC.

[0582] Patients with SD could be separated into two prognostic groups based on the CYFRA 21-1, CA 125 and CEA combination (score using an interaction term taking the histology associated performance of the biomarkers CYFRA 21-1 and CA 125 and optionally CEA into account), above or below median; high-risk and low-risk. The prognostic performance of the triple combination (score using an interaction term taking the histology dependent performance of the biomarkers CYFRA 21-1 and CA 125 and optionally CEA into account) increased when patients were separated based on the optimized cut-off, suggesting that patients with high-risk SD outcome are comparable to those with PD for OS, and patients with low-risk SD have outcome comparable to those with PR for OS and PFS.

[0583] We have demonstrated herein that CYFRA 21-1 and CA 125 absolute levels as such, a score combining the absolute levels of the two biomarkers and in particular a score taking into account CYFRA 21-1. CA 125 and CEA absolute levels can provide additional guidance to the first CT scan after the second treatment cycle in patients with indeterminate CT response, by differentiating those with SD into high and low risk groups for PFS and OS. Further, we have demonstrated that the biomarker score can be further improved by taking into account an interaction term based on tumor histology (SCC or adenocarcinoma) and optimizing the cut-off.

[0584] A particular surprising finding of this study was that the biomarker levels based on a single measurement after the first CT scan after the second treatment cycle showed a much better performance in the stratification of SD patients being at high risk for disease progression and SD patients being at low risk of disease progression than changes in the levels of the same biomarkers measured before treatment. This finding indicates that a single biomarker measurement without the cumbersome requirement of continuous monitoring of biomarker levels is sufficient and even best for predicting disease progression. Furthermore, the sufficiency of only a single measurement will open up this risk assessment also to patients that have not had a continuous biomarker survey and avoids any potential problems with usage of different methods for the measurements at different time points of the treatment.

[0585] These findings can help physicians to identify patients who are more or less likely to respond to therapy early (after Cycle 2). The ability to predict already at the time of the first CT scan in patients with SD may offer additional guidance in stable disease and thus indeterminate radiological results and help guiding further treatment decisions (e.g. monitoring for possible treatment adjustments).