Significance of Folate Receptor alpha and Thymidylate Synthase Protein Expression in Patients with Non-Small Cell Lung Cancer treated with Pemetrexed

Abstract

Introduction

Folate receptor alpha (FRA) regulates cellular uptake of folates and antifolates. Information about FRA protein expression in metastatic Non-Small Cell Lung Cancer (NSCLC) is limited. We investigated FRA as a biomarker for pemetrexed-based chemotherapy and compared it to thymidylate synthase (TS), the main target of pemetrexed.

Materials and Methods

Pre-treatment tumor specimens from 207 NSCLC patients with advanced disease were assessed for FRA and TS protein expression by immunohistochemistry using the H-score (range: 0–300) and correlated to patients’ clinicopathological data, radiographic response (RECIST), progression-free survival (PFS) and overall survival (OS).

Results

Low total (cytoplasmic and nuclear) TS protein expression (H-score <210) was associated with improved PFS (median: 5.6 vs. 3.5 months; HR=0.6379, P=0.0131) and prolonged OS (median: 22.5 vs. 11.5 months; HR: 0.5680, P=0.0107). An association between lower TS levels and response to pemetrexed-based therapy was found: mean H-score 187±5, median 180 for responders vs. 201±4, median 210, P=0.0244. High intracellular FRA expression (H-score ≥ 110) was associated with prolonged OS (28.9 vs. 11.7 months, HR=0.5316, P=0.0040) and a trend for association with PFS (5.6 vs. 4.1 months, HR=0.7395, P=0.0801) was noted. Membranous FRA expression was seen in 83% of patients, moreover, high membranous expression (H-score ≥20) was associated with improved PFS (5.6 vs. 3.7 months, HR=0.6445, P=0.0306) and OS (22.1 vs. 11.5 months, HR=0.5378, P=0.0131).

Conclusions

A large number of NSCLC patients have high expression of FRA and/or low level of TS expression. Expression levels of FRA and TS were associated with clinical benefit from pemetrexed-therapy.

INTRODUCTION

Lung cancer is the leading cause of cancer deaths worldwide and Non-Small Cell Lung Cancer (NSCLC) accounts for 80% of all lung cancer cases¹. The standard first-line therapy for patients with advanced NSCLC is a platinum-based doublet combination chemotherapy with or without bevacizumab². In clinical practice, pemetrexed, a multitarget antifolate drug, is used in combination with cisplatin³ or carboplatin⁴ in non-squamous cell carcinomas. Furthermore, it is administered as continuous maintenance therapy after platinum-based chemotherapy, as single agent after progression of first-line therapy^5,6 or it is given to patients who are medically unfit for platinum-based combination chemotherapy⁷. Pemetrexed primarily inhibits thymidylate synthase (TS). But at higher concentrations it also blocks folate-dependent enzymes such as dihydrofolate reductase (DHFR) and glycinamide ribonucleotide formyl transferase (GARFT), all of which are involved in the de novo biosynthesis of thymidine and purine nucleotides⁸.

Predictive biomarkers have proven to be effective for decisions regarding treatment with specific molecular targeted agents. However, for pemetrexed few studies have investigated predictive biomarkers for treatment efficacy and the majority of these have focused on TS itself. Previous studies have shown that low TS protein levels are associated with prolonged progression-free survival (PFS) of NSCLC patients undergoing pemetrexed-based treatment^9,10. However, inconsistent results have been reported regarding the association between objective response (OR), overall survival (OS) and TS expression levels^9,10. An association between improved OS and low TS levels was found in only one subgroup of adenocarcinoma patients⁹, but information about Caucasian patients is very limited.

For the transport of folates and antifolates into eukaryotic cells, three transporters have been identified: the ubiquitous reduced folate carrier (RFC), the proton-coupled folate transporter (PCFT), and a family of folate receptors of which folate receptor α (FRA) is the most widely studied^11,12. However, FRA-mediated transport contributes little to the uptake of either folates or antifolates unless the receptor is highly expressed relative to RFC, and/or transport mediated by RFC and other routes is impaired¹³. FRA is a glycosylphosphatidylinositol (GPI) anchored cell surface protein that is able to bind free folate with high affinity¹⁴ and assists in the uptake of antifolates via a classical mechanism of receptor-mediated endocytosis. Although FRA has a very limited expression profile in normal adult tissues, it is expressed in lung tissues¹⁵, in type I and II pneumocytes^16–20, basilar and mucociliary bronchiolar cells and bronchial epithelium^18–20.

High expression levels of FRA have been found in certain cancer cells of epithelial origin including certain lung tumors^19,21,22. Lung tumors that retain alveolar epithelial cell characteristics, such as adenocarcinoma, bronchioloalveolar carcinoma and large cell carcinoma, usually express detectable amounts of FRA protein^21,23.

Novel therapeutic agents directed against FRA have been developed within the last few years and are currently under clinical investigation for the treatment of lung cancer. These agents are either conjugates of a cytotoxic agent with folic acid (e.g. EC-145, Endocyte, Inc., West Lafayette, IN)²⁴ or antibodies eliciting robust antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) (e.g. farletuzumab, Morphotek, Inc., Exton, PA)²⁵. In vitro experiments showed that neither loss of constitutive FRA expression by RNA interference²⁶ nor the addition of a cytotoxic antibody (e.g. farletuzumab) significantly inhibited the transport of pemetrexed into NSCLC cells or interfered with the efficacy of pemetrexed²⁷.

Very limited information about FRA expression in metastatic NSCLC is available. This limits the development of potential novel drugs targeting this receptor. Early and locally advanced adenocarcinomas, large cell carcinomas and carcinoid tumors have been shown to express FRA protein, but protein expression of FRA in squamous cell carcinomas and small-cell carcinomas has rarely been observed^18,23,28. FRA gene expression as measured by real-time polymerase chain reaction (RT-PCR) in NSCLC patients with well-differentiated, early p-stage, pT1, pN0 cancers was shown to be significantly higher than those in poorly differentiated, advanced p-stage, pT2-4, pN1-3 patients²⁹. The disease-free survival (DFS) and 3-year survival rates of surgically treated patients with tumors expressing high FRA mRNA levels were significantly higher than those of patients with low FRA mRNA levels²⁹. In contrast, no associations between intracellular or membranous expression of FRA and DFS or OS were found in a study of 320 patients with early or locally advanced NSCLC²³. Recently, a significant association between improved OS and higher levels of membranous FRA expression in adenocarcinomas of the lung was reported²⁸.

Currently, there are no clinically validated biomarkers to predict response or outcome following pemetrexed-based chemotherapy. Therefore, we investigated the association of FRA and TS expression detected by immunohistochemistry (IHC) and clinical outcome following pemetrexed-based chemotherapy in a cohort of Caucasian patients with advanced NSCLC. Furthermore, we investigated the expression patterns of FRA and TS in advanced NSCLC and its correlation with clinicopathological parameters.

MATERIALS AND METHODS

Study Population

The study included all patients with NSCLC, who received pemetrexed-based chemotherapy at any chemotherapy line at the West German Cancer Center between November 2004 and October 2011 for the treatment of their metastatic disease. Analysis of tumor samples, which were available following routine diagnostic histopathological workup, provided adequate pre-treatment biopsies from 207 patients (Table 1). Clinicopathological data included age, gender, histology, complete history and radiographic findings (detailed information about the distribution of N-stages for the different histological subtypes is shown in Supp. Table 1). Tumor staging was based on the tumor, node, and metastasis (TNM) staging system (7^th edition) as proposed by the International Association for the Study of Lung Cancer (IASLC)³⁰. Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1³¹ was applied for the evaluation of response of NSCLC patients to treatment. PFS was calculated from the first day of chemotherapy treatment until progression or the last visit where a patient was alive without progression. OS was defined as the time between the start of pemetrexed-based chemotherapy until the date of death, or the date of last follow-up. Patients were censored at the last follow-up if still alive or lost to follow-up. Surveillance of PFS and OS for this study was stopped on February 1, 2012. Living patients provided written informed consent for the tissue used in this study. The study was approved by the Ethics Committee of the Medical Faculty of the University Duisburg-Essen (no. 10-4404).

Table 1.

Clinicopathological data of NSCLC patients: CR: complete remission; IASLC: International Association for the Study of Lung Cancer; OS: overall survival; PD: progressive disease; PFS: progression-free survival; PR: partial remission; and SD: stable disease. Note that percentages may not total 100 because of rounding.

Epidemiologic and clinicopathological data	Number of patients (%)
	207 (100%)
Age
Median age at diagnosis in years (range)	59 (23–90)
Gender
Female	94 (45%)
Male	113 (55%)
T-stage
T1	24 (12%)
T2	79 (38%)
T3	37 (18%)
T4	67 (32%)
N-stage
N0	38 (18%)
N1	18 (9%)
N2	81 (39%)
N3	70 (34%)
M-stage
M0	80 (39%)
M1	127 (61%)
IASLC-stage
I	8 (4%)
II	14 (7%)
III	58 (28%)
IV	127 (61%)
Histologic subtype
Adenocarcinoma	156 (75%)
Large cell carcinoma	12 (6%)
Squamous cell carcinoma	12 (6%)
Adenocarcinoma with large cell carcinoma	5 (2%)
Adenocarcinoma with squamous cell carcinoma	5 (2%)
Other combined tumors	10 (5%)
Other carcinomas	4 (2%)
Carcinoma not otherwise specified	3 (1%)
Histological grading
G1	8 (4%)
G2	63 (30%)
G3	130 (63%)
G4	3 (1%)
Unspecified	3 (1%)
Characteristics	Number of patients (%)
	207 (100%)
Chemotherapy
Pemetrexed plus cisplatin	69 (33%)
Pemetrexed plus carboplatin	31 (15%)
Pemetrexed plus cisplatin and carboplatin	2 (<1%)
Pemetrexed plus other compounds	5 (2%)
Pemetrexed single-agent	100 (48%)
Median number of cycles (range)	4 (1–23)
Median cumulative pemetrexed dose in mg (range)	3,235 (600–26,080)
Median duration of pemetrexed-based therapy in months (range)	2.8 (0.7–18.9)
Chemotherapy line
1st line	77 (37%)
2nd line	59 (29%)
3rd line	37 (18%)
4th line	20 (10%)
>4th line	14 (7%)
Response to pemetrexed-based treatment
CR	8 (4%)
PR	68 (33%)
SD	72 (35%)
PD	39 (19%)
Not evaluable	20 (10%)
Survival time from start of treatment to progression or death
Median PFS time in months (95% CI)	5.3 (4.7–6.0)
Median OS time in months (95% CI)	16.7 (10.9–22.4)

Tissue micro array (TMA) preparation

Sections were prepared from each formalin-fixed, paraffin-embedded (FFPE) tissue block and stained with hematoxylin and eosin for histological examination to identify 3 tumor-rich foci, which were marked with a cytology pen. A tissue arrayer (Beecher Instruments, Inc., Sun Prairie, WI, USA) was used to construct the TMA blocks. Three core samples of 1 mm size from marked tumor-rich regions of each donor FFPE block were sampled and transferred to a recipient paraffin block.

Immunohistochemistry (IHC) for FRA and TS

The TMA blocks were sectioned at 4 μm thick and mounted on PlusGOLD® slides (Thermo Scientific Fisher, Inc., Pittsburgh, PA, USA) for IHC staining. In cases of cores lost from TMA sections or inadequate tumor samples for TMA preparation, whole tissue sections were used.

The IHC detection of FRA protein has been recently described elsewhere²⁸. Briefly, after deparaffinization, rehydration and heat-induced antigen-retrieval, IHC staining process was done using Dako Autostainer Universal Staining System (Dako, Carpentaria, CA, USA). Tissue was blocked for endogenous peroxidase activity. Unconjugated antibody, mouse monoclonal FRA (clone 26B3¹⁴, dilution 2.5 μg/ml, Morphotek, Inc., Exton, PA, USA), was applied and incubated for 60 minutes at room temperature (RT). For negative control slide non-immunized unconjugated antibody, universal negative control mouse antibody (Dako), was applied and incubated under the same conditions. MACH4 Mouse Probe Primary Antibody Enhancer (Biocare Medical, Concord, CA, USA) was applied and incubated for 15 minutes followed by Polymer-HRP reagent (Biocare) for 20 minutes at RT. Visual color was developed with 3-3′-diaminobenzidine (DAB) tetra-hydrochloride (Dako). Slides were counterstained with Hematoxylin (Dako).

As FRA is expressed in the proximal tubules of the kidney, human kidney sections were used as a positive control for FRA IHC^23,28. Figure 1A shows highly specific staining of the FRA antibody, since FRA expression is restricted to the luminal surface of proximal tubule cells^17,19.

Figure 1.

IHC for folate receptor alpha (FRA) and thymidylate synthase (TS) in Non-Small Cell Lung Cancer (NSCLC) samples with different protein expression levels. A–E: IHC with an antibody against FRA (dilution 2.5 μg/ml). A) renal tissue was used as positive control, B) negatively, C) weakly, D) moderately, and E) strongly stained NSCLC sample. F–I: IHC with an antibody against TS (dilution 50 μg/ml). F) colorectal carcinoma was used as positive control; G) weakly, H) moderately, and I) strongly stained NSCLC sample. Magnification 1:400.

We have recently published the IHC procedure for TS³². Briefly, after baking the slides in dry oven for an hour, heat-induced antigen retrieval, and deparaffinization, the IHC staining procedure was performed using the Benchmark XT autostainer (Ventana, Tucson, AZ, USA). Endogenous peroxidase activity was blocked and unconjugated mouse anti-TS antibody (clone 4H4B1, dilution 50 μg/ml, Invitrogen, Frederick, MD, USA) was manually applied and incubated for 1 hour at RT. Tissue sections of colorectal carcinoma were included as controls, and lymphocytes were used as an intra-specimen positive control. For negative control, unconjugated non-immunized ready-to-use mouse antibody (Ventana) was applied as the primary antibody. Staining visualization was accomplished with ultraView Universal DAB detection kit (Ventana), post counterstaining with Bluing Reagent (Ventana).

IHC evaluations were performed independently by a pathologist (B.R.A) and a trained reader (D.C.C.). Ambiguous cases were reevaluated jointly until a consensus was reached. Furthermore, we performed a round-robin test on a subset of patients with the Laboratory Corporation of America (Los Angeles, CA, USA) revealing highly similar H-scores. Immunoreactivity levels in each section were assessed under a light microscope and images were captured at 400-fold magnification. Tumor staining intensity was graded on a scale from 0+ to 3+ and within each intensity category the percentage of tumor cells was noted. The percentage score was then multiplied by its intensity category to obtain a final “H-score”, ranging from 0 to 300^23,33. The highest H-score of the triplicate scores per patient was used for further analyses³². Tumor samples representing different H-scores for either FRA or TS are shown in Figure 1.

Evaluation of total TS protein expression was based on nuclear and cytoplasmic staining^9,33,34. For FRA, varied staining intensities and patterns of both the intracellular and membranous compartments were observed between cases^22,23. While FRA is a GPI-anchored receptor that cycles into the cell via an endocytic mechanism, remaining membrane associated, we decided to evaluate FRA staining for both the intracellular and membrane compartments separately.

Statistics

The objective of this study was to assess the impact of the two selected biomarkers (FRA and TS) on treatment response, and the outcome of patients with NSCLC. Another goal was to evaluate the intracellular and membranous FRA expression in metastatic NSCLC specimens to determine their expression pattern in different histological subtypes. Descriptive analyses were performed on these H-scores and correlated with patient’s clinical data: age at diagnosis, gender, histological grade or subtype (adenocarcinoma vs. large cell carcinoma vs. squamous cell carcinoma), and TNM/IASLC stage. A Spearman’s test³⁵ was used to further correlate as continuous variables intracellular and membranous FRA H-scores, total TS H-score, and age at diagnosis. Distributional differences of H-scores between different levels of categorical variables (response, gender, histological grade, subtype, or stage) were analyzed using either the Kruskal-Wallis test or Wilcoxon ranked sum test (equivalent to Kruskal-Wallis test when comparing two groups)³⁶. Receiver operating characteristic (ROC) curves³⁷ were constructed by calculating the sensitivity and specificity at several cutoff-values. The optimal cutoff-value was then given by the maximum of the Youden index³⁸.

For the Kaplan-Meier (KM) curves of PFS and OS, we tested binary H-score cutoff-values of biomarkers and applied the maximum chi-square method^28,39,40. Univariate and multivariate analyses of both PFS and OS were performed using the Cox proportional hazards model. All variables with P-values less than 0.10 at the time of univariate regression analysis were entered into the multivariate analysis model. For the assessment between survival and combined expression of FRA and TS, the H-score of total FRA expression (intracellular plus membranous expression) was determined by calculating the mean of intracellular and membranous expression. Furthermore, subgroup analyses for patients with first-line treatment and for patients with adenocarcinomas were calculated. A P-value less than 0.05 was considered statistically significant. Nevertheless, as this is a retrospective study all P-values or results of statistical tests should be regarded as exploratory. Statistical analyses were performed by using GraphPad Prism (Version 5.00 for Windows, GraphPad Software, SanDiego, CA, USA) or SPSS software package (Version 20, SPSS Inc., Chicago, IL, USA).

RESULTS

Patient cohort

Demographic and clinicopathologic data of the patients are shown in Table 1. Median PFS and OS of the entire cohort were 5.3 months (95%-CI: 4.7–6.0) and 16.7 months (95%-CI: 10.9–22.4), respectively. The main histologic subtype in the present cohort was adenocarcinoma, occurring in 156 patients (75%); large cell carcinoma and squamous cell carcinoma subtypes were each found in 12 patients (6%). Median PFS and OS of adenocarcinoma patients were 5.4 months (95%-CI: 4.6–6.2) and 16.7 months (95%-CI: 10.1–23.2), respectively. The majority of patients were males (n=113, 55%). 77 (37%) patients received pemetrexed-based first-line chemotherapy and in this subgroup median PFS was 7.3 months (95%-CI: 5.4–9.2) and median OS was 39.6 months (95%-CI: 19.4–59.9).

Intratumoral comparison of FRA and TS expression

Due to technical reasons, IHC and H-scoring was only successful for FRA in 207 cases, and 196 cases for TS. Intracellular expression of FRA was strongly correlated with membranous expression (Figure 2A, Spearman’s correlation; r=0.8460; P<0.0001). Similarly, strong correlations in the subgroups of adenocarcinoma or large cell carcinoma patients were found (Suppl. Figures 1A/B). A weaker, but still significant correlation in squamous cell carcinomas was observed (r=0.6597, P=0.0196; Suppl. Figure 1C). There were no correlations between TS protein expression and intracellular (r=−0.1029, P=0.1513) or membranous FRA expression (r=−0.1227; P=0.0868).

Figure 2.

A) Intratumoral correlation of intracellular and membranous protein expression of FRA. Scatter-plots of (B) TS, (C) intracellular, and (D) membranous FRA protein expression in different histological subtypes (vertical line: median). ADC: adenocarcinoma, LCC: large cell carcinoma, SCC: squamous cell carcinoma.

Since data on FRA expression in metastatic NSCLC are limited, we were particularly interested in the incidence of intracellular and membranous FRA expression. Intracellular or membranous FRA expression (H-score higher than 5) were detected in 192 patients (93%) or 171 patients (83%), respectively. As an H-score of ≥ 5 is a low cutoff-value, we tested higher cutoff-values as well. Considering an H-score of ≥ 100, 140 patients (68%) or 99 patients (48%) had tumors with high intracellular or membranous expression. Using an H-score of ≥ 200 as cutoff-value to distinguish between low and high FRA expressing tumors 42 patients (20%) or 30 patients (15%) showed tumors with strong intracellular or membranous expression. To summarize, the incidence of FRA membrane expression observed in the present study compares favorably with a recent report²⁸.

Association between clinicopathological data and TS expression

The median H-score for total TS expression was 200 (range: 60–290). There was no statistically significant correlation between age, sex, histological grades or subtypes (Figure 2B), T- or M-stages and total TS protein levels (Table 2A). N-stages were associated with total TS protein expression (P_KW=0.0003). Interestingly, lower total TS protein expression was associated with early N-stages at diagnosis (mean for N0/N1: 174±6, median 160 vs. mean for N2/N3: 203±3, median 207.5) (P<0.0001). Lower total TS protein levels were also found in early IASLC stages (mean for IASLC stages I+II: 176±11, median 160) compared to advanced IASLC stages (mean for IASLC stages III + IV: 198±3, median 200; P=0.0331).

Table 2.

Associations between clinicopathological data and (A) total TS protein expression, (B) intracellular or (C) membranous protein expression of FRA. IASLC: International Association for the Study of Lung Cancer.

Table 2A
Characteristic	Number of patients (%)	Mean total TS H-score ± SEM, median	P-value
Histological subtype			PKW=0.9902
Adenocarcinoma	147 (75%)	196 ± 3, 190
Large cell carcinoma	12 (6%)	184 ± 17, 202.5
Squamous cell carcinoma	10 (5%)	191 ± 16, 205
Histological subtype			P=0.9856
Adenocarcinoma	147 (75%)	196 ± 3, 190
Squamous cell carcinoma	10 (5%)	191 ± 16, 205
Gender			P=0.5982
Female	90 (46%)	196 ± 5, 200
Male	106 (54%)	196 ± 4, 200
Grading			PKW=0.5688
G1	7 (4%)	213 ± 12, 210
G2	58 (30%)	192 ± 5, 190
G3	126 (65%)	196 ± 4, 200
G4	3 (2%)	207 ± 7, 200
Grading			P=0.4339
G2	58 (30%)	192 ± 5, 190
G3	126 (65%)	196 ± 4, 200
T-stage			PKW=0.3635
T1	24 (12%)	190 ± 8, 190
T2	74 (38%)	200 ± 5, 200
T3	35 (18%)	187 ± 7, 190
T4	63 (32%)	196 ± 5, 200
N-stage			PKW=0.0003
N0	37 (19%)	173 ± 7, 160
N1	17 (9%)	176 ± 12, 160
N2	75 (38%)	205 ± 4, 220
N3	67 (34%)	201 ± 5, 205
N-stage			P<0.0001
N0 + N1	54 (28%)	174 ± 6, 160
N2 + N3	142 (72%)	203 ± 3, 207.5
M-stage
M0	76 (39%)	193 ± 5, 200	P=0.7362
M1	120 (61%)	197 ± 4, 200
IASLC-Stage			PKW=0.1586
IASLC I	8 (4%)	174 ± 15, 160
IASLC II	13 (7%)	178 ± 16, 160
IASLC III	55 (28%)	199 ± 6, 200
IASLC IV	120 (61%)	197 ± 4, 200
IASLC-Stage			P=0.0331
IASLC I + II	22 (11%)	176 ± 11, 160
IASLC III + IV	185 (89%)	198 ± 3, 200

Table 2B
Characteristic	Number of patients (%)	Mean intracellular FRA H-score ± SEM, median	P-value
Histological subtype			PKW=0.0368
Adenocarcinoma	156 (77%)	134 ± 6, 130
Large cell carcinoma	12 (6%)	92 ± 23, 67.5
Squamous cell carcinoma	12 (6%)	87 ± 19, 90
Histological subtype			P=0.0386
Adenocarcinoma	156 (77%)	134 ± 6, 130
Squamous cell carcinoma	12 (6%)	87 ± 19, 90
Gender			P=0.2520
Female	94 (45%)	134 ± 8, 135
Male	113 (55%)	122 ± 7, 110
Grading			PKW=0.0318
G1	8 (4%)	105 ± 25, 105
G2	63 (30%)	151 ± 9, 155
G3	130 (63%)	121 ± 7, 110
G4	3 (1%)	57 ± 32, 30
Grading			P=0.0173
G2	63 (30%)	151 ± 9, 155
G3	130 (63%)	121 ± 7, 110
T-stage			PKW=0.3841
T1	24 (12%)	147 ± 13, 157.5
T2	79 (38%)	129 ± 9, 125
T3	37 (18%)	130 ± 12, 120
T4	67 (33%)	118 ± 10, 110
N-stage			PKW=0.0120
N0	38 (18%)	148 ± 11, 152.5
N1	18 (9%)	167 ± 16, 172.5
N2	81 (39%)	114 ± 9, 100
N3	70 (34%)	123 ± 9, 110
N-stage			P=0.0014
N0 + N1	56 (27%)	154 ± 9, 110
N2 + N3	151 (73%)	118 ± 6, 70
M-stage
M0	80 (39%)	135 ± 8, 132.5	P=0.1916
M1	127 (61%)	123 ± 7, 110
IASLC-Stage			PKW=0.3385
IASLC I	8 (4%)	153 ± 16, 150
IASLC II	14 (7%)	146 ± 22, 150
IASLC III	58 (28%)	131 ± 10, 125
IASLC IV	127 (61%)	123 ± 7, 110
IASLC-Stage			P=0.0848
IASLC I + II	22 (11%)	153 ± 14, 150
IASLC III + IV	185 (89%)	125 ± 6, 120

Table 2C
Characteristic	Number of patients (%)	Mean membranous FRA H-score ± SEM, median	P-value
Histological subtype			PKW=0.0059
Adenocarcinoma	156 (77%)	103 ± 7, 97.5
Large cell carcinoma	12 (6%)	59 ± 20, 30
Squamous cell carcinoma	12 (6%)	46 ± 24, 0
Histological subtype			P=0.0066
Adenocarcinoma	156 (77%)	103 ± 7, 97.5
Squamous cell carcinoma	12 (6%)	46 ± 24, 0
Gender			P=0.0325
Female	94 (45%)	108 ± 8, 120
Male	113 (55%)	86 ± 8, 60
Grading			PKW=0.0375
G1	8 (4%)	66 ± 25, 55
G2	63 (30%)	116 ± 11, 130
G3	130 (63%)	91 ± 7, 70
G4	3 (1%)	20 ± 20, 0
Grading			P=0.0674
G2	63 (30%)	116 ± 11, 130
G3	130 (63%)	91 ± 7, 70
T-stage			PKW=0.3585
T1	24 (12%)	118 ± 15, 135
T2	79 (38%)	98 ± 9, 90
T3	37 (18%)	96 ± 13, 105
T4	67 (33%)	86 ± 10, 60
N-stage			PKW=0.1770
N0	38 (18%)	106 ± 12, 120
N1	18 (9%)	129 ± 19, 140
N2	81 (39%)	89 ± 9, 70
N3	70 (34%)	90 ± 9, 67.5
N-stage			P=0.0398
N0 + N1	56 (27%)	114 ± 10, 130
N2 + N3	151 (73%)	89 ± 7, 70
M-stage
M0	80 (39%)	105 ± 9, 105	P=0.1916
M1	127 (61%)	90 ± 7, 70
IASLC-Stage			PKW=0.4508
IASLC I	8 (4%)	78 ± 27, 80
IASLC II	14 (7%)	111 ± 23, 135
IASLC III	58 (28%)	107 ± 11, 100
IASLC IV	127 (61%)	90 ± 7, 70
IASLC-Stage			P=0.9624
IASLC I + II	22 (11%)	100 ± 17, 125
IASLC III + IV	185 (89%)	95 ± 6, 90

TS expression is associated with treatment response

Low total TS expression was associated with objective response (OR) (mean 187±5, median 180 for responders (CR+PR) vs. 201±4, median 210 for non-responders (SD+PD), P=0.0244) (Suppl. Table 2A), but not disease control (DC) meaning objective response or disease stabilization (mean 194±4, median 190 for CR/PR/SD vs. 201±7, median 210 for PD, P=0.1375) (Suppl. Table 2A). As we observed a significant association between low total TS protein levels and OR, we calculated ROC curves analyzing TS and response. We discovered an area-under-curve (AUC) value of 0.5997 (P=0.0247). We used the optimal cutoff of 197.5 before estimating the sensitivity, specificity, positive and negative predictive value for response, which were 62%, 60%, 70%, and 52%, respectively.

Association between survival and TS expression

In the univariate analysis considering TS, low total TS expressers (H-score <210, χ²=6.151) had a significantly longer PFS (5.6 vs. 3.5 months) as compared to high expressers (P=0.0131; HR=0.6379, 95%-CI: 0.4472–0.9100) (Figure 3A). A similar association between TS and OS was observed: low TS expressers (H-score <210, χ²=6.516) had a prolonged OS (22.5 vs. 11.5 months) as compared to high expressers (P=0.0107; HR=0.5680, 95%-CI: 0.3678–0.8769) (Figure 3B). We performed a Cox proportional hazard model for PFS/OS and total TS protein levels. In the univariate analysis the association between PFS and total TS protein levels remained significant (P_COX=0.047). No significant associations between clinicopathological parameters and PFS were observed. The univariate analysis considering continuous TS protein levels showed a trend for an association between OS and TS (P_COX=0.077), but no association was observed after adjustment for histological grades and N-stages (P_COX=0.152). We performed survival analyses for the subgroups of patients who received first-line treatment or suffered from adenocarcinomas (Suppl. Figures 2A/B and 3A/B). Both revealed similarly significant results.

Figure 3.

Analyses of total thymidylate synthase (TS) protein expression and (A) progression-free survival (PFS) or (B) overall survival (OS); analyses of intracellular protein expression of folate receptor alpha (FRA) and (C) PFS or (D) OS; analyses of membranous FRA protein expression and (E) PFS or (F) OS.

Association between clinicopathological data, response and intracellular FRA expression

The median H-score for intracellular FRA expression was 120 (range: 0–290). Intracellular expression was significantly higher in adenocarcinomas (mean 134±6, median 130) compared to large cell carcinomas (mean 92±23, median 67.5) or squamous cell carcinomas (mean 87±19, median 90) (P_KW=0.0368) (Figure 2C). Differentiation of the tumors was significantly associated with intracellular expression (P_KW=0.0318) (Table 2B). Higher intracellular levels were found in moderately differentiated tumors (G2: mean 151±9; median 155) compared to poorly differentiated tumors (G3: mean 121±7, median 110; P=0.0173). T- or M-stages were not associated with intracellular expression. But an association between N-stages and intracellular expression was found (P_KW=0.0120). Of note, higher intracellular levels were observed in N0/N1-stages (mean H-score: 154±9, median 110) compared to N2/N3-stages (mean H-score 118±6, median 70, P=0.0014). No association between response and intracellular FRA expression levels was noted.

Association between survival and intracellular FRA expression

A trend for a significant association between high intracellular FRA expression (H-score ≥ 110, χ²=3.063) and prolonged PFS (5.6 vs. 4.1 months; P=0.0801; HR=0.7395, 95%-CI: 0.5275–1.037) was observed (Figure 3C). High expressers of intracellular FRA (H-score ≥ 110, χ²=8.272) were associated with an improved OS (28.9 vs. 11.7 months) as compared to low expressers (P=0.0040; HR=0.5316, 95%-CI: 0.3456–0.8177) (Figure 3D). We performed a Cox proportional hazard model for OS and intracellular FRA protein levels. In the univariate analysis the association between OS and intracellular protein levels was significant (P_COX=0.021), but when adjusting for histological grades and N-stages in the multivariate analysis, there was a trend for an association (P_COX=0.056). Survival analyses for the subgroups of first-line treatment or adenocarcinoma patients (Suppl. Figures 2C/D and 3C/D) showed that high expressers had an improved OS (P=0.0203 or P=0.0028, respectively). However, no significant associations between intracellular FRA expression and PFS were found in theses subgroups (P=0.1224 or P=0.1759, respectively).

Association between clinicopathological data, response and membranous FRA expression

The median H-score for membranous FRA expression was 90 (range: 0–290). FRA was significantly more expressed in adenocarcinomas (mean 103±7, median 97.5) compared to large cell carcinomas (mean 59±20, median 30) or squamous cell carcinomas (mean 46±20, median 0) (P_KW=0.0059) (Figure 2D). Membranous expression was significantly higher in females (mean 108±8, median 120) than in males (mean 86±8, median 60) (P=0.0325) and significantly associated with differentiation of the tumors (P_KW=0.0375). The difference between membranous levels of moderately differentiated carcinomas (G2: mean H-score 116±11; median 130) and poorly differentiated carcinomas (G3: mean H-score 91±7, median 70) showed a trend for a significant association (P=0.0674). T- or M-stages were not associated with membranous expression (Table 2C). Higher membranous levels were observed in N0/N1-stages (mean 114±10, median 130) compared to N2/N3-stages (mean 89±7, median 70, P=0.0398). No association between response and membranous FRA expression levels was observed.

Association between survival and membranous FRA expression

High membranous FRA expression (H-score ≥ 20, χ²=4.676) was associated with a prolonged PFS (5.6 vs. 3.7 months, P=0.0306, HR=0.6445; 95%-CI: 0.4328–0.9597; Figure 3E). Similarly, high expressers (H-score ≥ 20, χ²=6.156) showed an improved OS (22.1 vs. 11.5 months, P=0.0131, HR=0.5378, 95%-CI: 0.3294–0.8778; Figure 3F). Again, we performed survival analyses for the subgroups of first-line treatment or adenocarcinoma patients (Suppl. Figures 2E/F and 3E/F). Strong associations between prolonged OS and high membranous FRA expression were seen in both of these subgroups (P<0.0001 or P=0.0064, respectively). There was a trend for an association between high membranous expression and improved PFS in the subgroup of first-line treatment patients (P=0.0514), and the association was significant in the subgroup of adenocarcinoma patients (P=0.0423).

Analyses of survival and combined TS/FRA expression

We performed survival analyses considering the combined total FRA and TS expression (Figure 4). PFS of patients with tumors expressing high levels of total FRA and low total TS levels was prolonged (5.8 months) compared to patients with low total FRA and high total TS levels expressing tumors (3.1 months; P=0.0033; HR=0.4071, 95%-CI: 0.2235–0.7414; Figure 4C). Similarly, patients with tumors expressing high levels of total FRA and low total TS levels had a significant improvement to OS of 28.9 months compared to patients with tumors showing low levels of total FRA and high total TS levels (6.9 months; P=0.0045; HR=0.2133, 95%-CI: 0.07341–0.6198; Figure 4D) or compared to patients with tumors containing high levels of both total FRA and total TS (15.2 months; P=0.0205, HR= 0.5650, 95%-CI: 0.3486–0.9157; Figure 4F).

Figure 4.

Analysis of the combination of total folate receptor alpha (FRA) and thymidylate synthase (TS) protein expression and (A) progression-free survival (PFS) or (B) overall survival (OS); comparison of survival times between high total FRA plus low total TS expressers and low total FRA plus high total TS-expressers (C/D) or high total FRA plus high total TS-expressers (E/F). For the categorization of low expressers or high expressers the following H-scores were used: total TS: 210 (PFS χ²=6.151, OS χ²: 6.516), total FRA: 60 (PFS χ²=4.032) or 20 (OS χ ²=5.541).

DISCUSSION

Pemetrexed is a multitarget antifolate drug approved for the treatment of advanced NSCLC. While the mechanism for the drug activity is known, no predictive marker has so far been validated for better selection of patients, in contrast to other targeted treatments. The identification of predictive markers for effective therapy would be helpful for maximizing therapeutic efficacy and minimizing non-effective treatment in patients with cancer. Therefore, novel strategies for the prediction of clinical response for known antitumor agents, such as pemetrexed, are urgently required for individualized therapy.

In the current study we confirmed an association between OR and low total TS protein expression (P=0.0244), which was shown recently¹⁰. In our study, low TS-expressers had a median PFS time of 5.6 months compared to 3.5 months of high TS-expressers (P=0.0131). Similarly, median OS time was improved in low TS-expressers (22.5 months) compared to high TS-expressers (11.5 months, P=0.0107). Our results are partly in accordance with other reports^9,10, as we observed an association between median PFS or OS times and TS expression levels. However, in the aforementioned studies, OS was only associated with TS levels in patients with adenocarcinomas⁹. Surprisingly, we did not find any association between TS protein levels and the histological subtypes of adenocarcinoma, large cell carcinoma, and squamous cell carcinoma. Some studies reported higher TS mRNA levels in squamous cell carcinomas compared to adenocarcinomas of the lung^41,42. However, our results are consistent with another study using automated protein in situ quantification (AQUA)⁴³ and our recently published study³², although these studies^32,43 investigated TS protein expression in early and locally-advanced NSCLC. Furthermore, it should be mentioned that the percentages of large cell or squamous cell carcinomas in our study is quite low (6% each, respectively), thus limiting the comparison of FRA and TS expression between the different histological subtypes. These patients received pemetrexed until this drug was exclusively approved for metastatic non-squamous NSCLC patients in 2009. Although we found significant associations between OR, PFS, OS, and TS protein expression, it should be noted that the AUC value of the ROC curve predicting response (0.5997) does not appear to support the idea that TS is a good predictive biomarker.

No associations between response and intracellular or membranous FRA levels were found, neither intracellular nor membranous FRA expression are robust enough for a predictive biomarker. These results may not be unexpected as a study investigating FRA in Malignant Pleural Mesothelioma did not reveal an association between total FRA protein expression and response to pemetrexed-based chemotherapy⁴⁴. Additionally, in the latter study, no associations between time-to-treatment failure or OS and total FRA expression were noted. However, in our study we investigated FRA expression in metastatic NSCLC in the intracellular and membrane compartments separately and found associations between high intracellular or membranous FRA expression and improved OS (P=0.0040 or P=0.0131, respectively). Only high membranous FRA expression was associated with a prolonged PFS (5.6 vs. 3.7 months, P=0.0306). Yet due to the lack of a control group (cohort of patients with pulmonary adenocarcinomas without pemetrexed-based treatment) we were unable to evaluate the prognostic value of intracellular or membranous FRA expression.

In our study of NSCLC patients with advanced disease, higher intracellular or membranous FRA levels were observed in more differentiated tumors. Assessment of FRA gene expression in early and locally advanced stages of NSCLC (n=117) revealed similar results²⁹. No associations between intracellular or membranous FRA expression and T- or M-stages were observed. But early N-stages were associated with higher expression levels of this protein in our series of metastatic NSCLC, consistent with results from another study with early and locally advanced NSCLC patients²⁹.

We examined the incidence of membranous FRA expression in metastatic NSCLC. Levels of membranous FRA expression (H-score ≥ 5) was detected in 171 patients (83%). But this H-score is a low cutoff-value, and we tested higher cutoff-values in addition such as an H-score of ≥ 100 revealing that about half of the patients (n=99, 48%) had tumors with higher membranous FRA levels. These results indicate that FRA is a potential target for novel therapeutics in metastatic NSCLC and cytotoxic-conjugated antibodies^45,46 as well as antibodies inducing ADCC and CDCC²⁵ are being tested in clinical trials. We found significant differences in FRA expression between adenocarcinomas and squamous cell carcinomas. Our results are consistent with other studies, in which FRA was predominantly expressed in adenocarcinomas, relative to squamous cell carcinomas^18,28. However, it should be mentioned that in one of these studies the M-score was used, which is another semi-quantitative assessment of FRA protein expression²⁸. To summarize, novel FRA-directed therapeutics should primarily be investigated in adenocarcinomas of the lung with greater differentiation.

We noted that the median H-score of intracellular FRA expression was higher than the median H-score of membranous expression, which is consistent with a recently published study²³. However, the significance of intracellular FRA expression is unclear. The intracellular staining is most likely a combination of several things: a) some background staining in the assay; b) a result of the process of tissue preparation for IHC (as GPI-anchored protein, FRA is susceptible to leaching or stripping from the membrane during alcohol washes); c) FRA expression in the cytoplasm where FRA is synthesized in the endoplasmic reticulum⁴⁷; and d) membrane-bound FRA undergoes an endocytic pathway⁴⁸, in which clathrin-coated buds and vesicles and endosomal vacuoles are loaded with the receptor and transported intracellularly. However, only the membranous form of FRA is biologically active/relevant and transports folates and antifolates. Therefore, the associations between intracellular FRA expression and OS should be interpreted with caution.

While the current study demonstrates some encouraging results, there are several limitations. Firstly, this is a retrospective study. Only patients with available pre-treatment tumor specimens were included in our study. Secondly, the clinical response and survival data are based on a retrospective analysis of medical records and image readings and are not collected from a prospective clinical trial. Various quantitative and semi-quantitative methods have been employed in different studies for measuring FRA expression in tissue biopsies, including IHC, radioimmunoassays, quantitative autoradiography, and cytofluorimetric analysis (all of which use anti-FR antibodies) as well as quantitative RT-PCR, fluorescence in situ hybridization, and radioligand binding assays¹⁹. However, the applicability of many of those methods to paraffin-embedded tissues is limited⁴⁹. Unfortunately, results from antibody-based methods can be hampered by factors such as non-specific binding to tissues, inaccessibility to the antigen, heterogenous antigen expression and/or poor tissue sampling, and the possibility that the antibody might not distinguish between functional FRA and non-functional FRA, both of which have been demonstrated in human cells¹⁹.

In conclusion, in NSCLC patients treated with pemetrexed-based chemotherapy, low total TS protein expression is associated with a better OR and an improved PFS/OS. For FRA, associations between prolonged PFS/OS and membranous expression were observed. Membrane-bound FRA was detected in a large percentage of NSCLC-patients indicating a clear target of FRA by novel therapeutics as a favorable goal. Future prospective studies regarding FRA and TS are needed for further validations.

Supplementary Material

2

Suppl. Table 1: Distribution of N-stages for the different histological subtypes of adenocarcinoma, large cell carcinoma, squamous cell carcinoma, exclusive of other histological subtypes. In general, the majority of the patients had higher N-stages with no significant differences between the histological subtypes being observed.

Suppl. Table 2: Association between protein expression of total TS and objective response (OR) or clinical disease control (DC). (A) Responding and non-responding patients of the entire cohort were compared and significant differences in the mean or median TS H-score for OR were found. (B) Comparison of patients who received first-line treatment revealed a trend for associations between the mean or median TS H-scores and OR or DC. (C) Patients with adenocarcinomas were compared and significant differences in the mean or median TS H-score for OR were observed.

3. Suppl. Figure 1.

Correlation between intracellular and membranous protein expression of folate receptor alpha (FRA) in different histological subtypes: (A) in adenocarcinomas (ADC), (B) in large cell carcinomas (LCC), (C) in squamous cell carcinomas (SCC), and (D) in all other histological subtypes (OTH).

4. Suppl. Figure 2.

Analysis of a subgroup of patients with first-line pemetrexed-based treatment. Analyses of total thymidylate synthase (TS) protein expression and (A) progression-free survival (PFS) or (B) overall survival (OS); analyses of intracellular protein expression of folate receptor alpha (FRA) and (C) PFS or (D) OS; analyses of membranous FRA protein expression and (E) PFS or (F) OS.

5. Suppl. Figure 3.

Analysis of a subgroup of patients with adenocarcinomas of the lung. Analyses of total thymidylate synthase (TS) protein expression and (A) progression-free survival (PFS) or (B) overall survival (OS); analyses of intracellular protein expression of folate receptor alpha (FRA) and (C) PFS or (D) OS; analyses of membranous FRA protein expression and (E) PFS or (F) OS.