CCTα is a novel antigen detected by the anti-ERCC1 antibody 8F1 with biomarker value in lung and head and neck squamous cell carcinomas

Abstract

BACKGROUND

Determination of in situ protein levels of ERCC1 with the antibody (8F1) is prognostic of survival in non-small-cell lung cancer (NSCLC). We previously demonstrated that 8F1 recognizes a second nuclear antigen. We identified this antigen and analyzed its value as a biomarker of clinical outcomes.

METHODS

The second antigen was identified by mass spectrometry. Protein identity and antibody specificity were confirmed through knockdown and overexpression experiments. Immunohistochemistry (IHC) of 187 early stage NSCLC samples and 60 head and neck squamous cell carcinomas (HNSCC) was used to examine the influence of the second antigen on 8F1 immunoreactivity and association with patient outcomes.

RESULTS

Cholinephosphate citidylyl transferase–α (CCTα, a.k.a. phosphate citidylyl transferase 1 choline alpha (PCYT1A), a phospholipid synthesis enzyme regulated by RAS, is the second antigen of 8F1. In NSCLC, CCTα contributed (rho = 0.38) to 8F1 immunoreactivity. In squamous cell carcinoma of the lung, CCTα was the dominant determinant of 8F1 immunoreactivity, while its contribution in other subtypes of lung cancer was negligible. High expression of CCTα, but not ERCC1, was prognostic of longer disease-free (log-rank p = 0.002), and overall survival (log-rank p = 0.056). Similarly, in HNSCC, CCTα contributed strongly to 8F1 immunoreactivity (rho = 0.74), and high CCTα expression was prognostic of survival (log-rank p = 0.022 for DFS and p = 0.027 for OS).

CONCLUSIONS

CCTα is the second antigen detected by 8F1. High CCTα expression is prognostic of survival in NSCLC treated by surgery alone and HNSCC. CCTα is a promising biomarker of patient survival and deserves further study.

INTRODUCTION

Platinum agents are the cornerstone of chemotherapy for non-small-cell lung cancer (NSCLC) and head and neck squamous cell carcinomas (HNSCC) (www.NCCN.org), but plagued with a high frequency of drug resistance and toxic side effects. The identification of molecular markers to predict therapeutic efficacy is a prerequisite for personalized cancer therapy and improving poor outcomes. Molecules involved in DNA damage repair are rational biomarker candidates. They are involved in maintaining genetic stability, impacting cancer initiation and progression, and response to genotoxic therapeutic agents. ERCC1-XPF is a bipartite endonuclease essential for repair of bulky DNA adducts and interstrand crosslinks, the types of damage caused by platinum agents.¹ In patients with early-stage NSCLC treated with surgery alone, high levels of ERCC1 protein had initially been reported to be associated with longer survival, presumably a result of maintaining genome stability.^{2, 3} However, in patients treated with genotoxic agents, low ERCC1 expression is associated with better survival,² presumably because of greater drug sensitivity as a result of reduced repair capacity. The relationship between ERCC1 and survival in NSCLC treated with DNA damaging agents was confirmed in phase III clinical trials^{2, 4} and meta-analyses.^5-7 In fact, ERCC1 was used as an enrichment biomarker in clinical trials until two recent reports raised doubts regarding its prognostic and predictive reliability.^{8, 9} Most studies that had established the predictive value of ERCC1 protein levels as a biomarker had used the monoclonal antibody 8F1.

We previously demonstrated that 8F1 recognizes a second nuclear protein.¹⁰ This cross-reactivity is important, as 8F1 was unable to discriminate between normal and ERCC1-XPF-deficient cells.¹¹ It is therefore possible that 8F1 signals reported in prior studies correspond to both proteins. Notably, in lung cancer, an XPF-specific antibody did not predict outcome, whereas 8F1 did.^{12, 13} In HNSCC, a specific ERCC1 antibody (FL297) predicted survival, while 8F1 did not.¹⁴ These observations suggest two hypotheses. First, ERCC1-specific antibodies may be more useful than 8F1 as they measure ERCC1 expression without noise introduced by cross-reactivity with other proteins. Second, the unidentified protein recognized by 8F1 may have intrinsic biomarker value that significantly contributed to results obtained in clinical trials that used 8F1.

Using mass spectrometry, we identified the second antigen of 8F1 as cholinephosphate citidylyl transferase–α (CCTα), a rate-limiting enzyme involved in the synthesis of phosphatidyl choline. We confirmed that CCTα is recognized by 8F1 and that this cross-reactivity significantly influences the 8F1 signal in clinical samples. We found that CCTα protein levels are a determinant of clinical outcomes in two independent cohorts of NSCLC and HNSCC.

MATERIALS AND METHODS

Antibodies and Chemicals

The antibodies used were anti-ERCC1 8F1 (Sigma-Aldrich), D-10 and FL297 (Santa Cruz Biotechnology), and EP2143Y (Abcam), anti-CCTα/PCYT1α (Sigma-Aldrich), and anti-β-tubulin (Sigma-Aldrich).

Cell Culture and Plasmids

Ovarian carcinoma A2780 cells and HeLa-S3 cells were cultured in RPMI-1640 medium with 10% FBS. C-terminal GFP tagged human CCTα (Origene) was transfected into HeLa-S3 cells, and genetycin was used to select stable clones. CCTα was knocked-down with shRNA constructs (Origene). Stable clones were selected with puromycin. ERCC1-deficient control cells were obtained from the Coriell Institute for Medical Research.¹⁵

Immunoprecipitation

Protein A/G beads were conjugated to 8F1 (non-specific) or D-10 (specific) antibody. A2780 cells were lysed in RIPA buffer with protease inhibitors and incubated with antibody-conjugated beads. The beads were boiled in Lameli buffer, and proteins were resolved by SDS-PAGE. The gel was silver-stained (BioRad), and bands were analyzed by mass spectrometry.

Immunohistochemistry and In Situ Quantification

Cell pellets were formalin fixed and paraffin embedded (FFPE). For tumor analysis, tissue microarrays of FFPE tumors and control tissues were used. Each tumor was represented by three distinct cores for NSCLC and two for HNSCC. Tissue microarrays were processed for antigen retrieval and IHC as described elsewhere.^{3, 15} The antibody dilutions used were 1:100 for 8F1, 1:250 for FL297, 1:200 for EP2143Y, and 1:200 for CCTα. The nuclear signal intensity was quantified by automated quantitative analysis for NSCLC³ and APERIO for HNSCC.¹⁵

Tumor Samples and Patient Cohorts

The well-characterized, prospectively collected, single-institution cohort of 187 early-stage NSCLC patients has been described elsewhere.³ Using the reverse Kaplan-Meier method, the median follow-up for overall (OS) and disease-free survival (DFS) was 69.0 and 35.9 months, respectively. The HNSCC cohort was a retrospective, single-institution cohort of 80 representative pretreatment biopsy or resection specimens described elsewhere.¹⁶ The median follow-up was 41 months. All sites in the aerodigestive tract, stages, and treatments were included. Demographic and clinical characteristics of both cohorts are summarized in Table 1. The study was approved by the IRBs of the respective institutions and conformed to the Helsinki declaration.

Table 1.

Demographic and Clinical Characteristics of Patients

Variable	Lung (n=187)	Head & Neck (n=60)
Age
Median Range (min, max)	70 (45, 84)	58 (24, 80)
Sex
Women	86 (46%)	17 (28%)
Men	101 (54%)	43 (72%)
Race
African American	6 (3%)	Data missing
Caucasian	179 (96%)
Unknown	2 (1%)
Smoking History
Non-Smoker	11 (6%)	12 (20%)
Smoker	162 (86%)	48 (80%)
Unknown	14 (7%)
Median Pack-years Range (min, max)	50 (0,300)	Data missing
PS
0	128 (68%)	Data missing
1	48 (26%)
Unknown	11 (6%)
Tumor Site and Type
Lung	187 (100%)
Oral cavity		25 (42%)
Oropharynx		13 (22%)
Hypopharynx		1 (2%)
Larynx		17 (28%)
H&N, not otherwise specified		4 (7%)
Primary	187 (100%)	49 (82%)
Recurrent	0 (0%)	11 (18%)
Stage
Lung Cancer
IA	85 (45%)
IB	102 (55%)
Head & Neck Cancer
I		3 (5%)
II		9 (15%)
III		6 (10%)
IV		31 (52%)
Unknown		11 (18%)
Histology
Adeno	96 (51%)	0 (0%)
Large Cell	23 (13%)	0 (0%)
Squamous Cell	68 (36%)	60 (100%)
Weight loss (>5% over 3 months)
Absent	159 (85%)	Data missing
Present	14 (7%)
Unknown	14 (7%)

Statistical Methods

The primary endpoint for analysis was DFS, defined as the time interval from the date of resection to disease recurrence or death. The secondary endpoint was OS, defined as the time interval from the date of resection to death from any cause. Detailed statistical methods are provided in Supplemental information. All p-values were two-sided with a significance level of 0.05. All calculations were performed with R Version 2.14.0 (R Project for Statistical Computing, Vienna, Austria).

RESULTS

8F1 Recognizes CCTα

We previously demonstrated that the monoclonal antibody 8F1 recognizes both ERCC1 and a second, unidentified nuclear protein that may interfere with the measurement of ERCC1 in clinical samples.¹⁰ To determine the identity of this protein, we immunoprecipitated whole cell lysates of A2780 cells using 8F1 or the ERCC1-specific monoclonal antibody D-10¹¹ and separated proteins by electrophoresis (Fig 1A); although the D-10 anitbody is ERCC1 specific by immunoblotting, it is not suitable for immunohistochemical detection of ERCC1. As expected, both antibodies pulled down ERCC1. The 8F1 antibody also precipitated a second, slower migrating protein (upper band), which could be seen by silver stain. Mass spectrometry identified the band as CCTα, a lipid-modifying enzyme involved in phosphatidyl choline synthesis¹⁷ (Fig 1B) that localizes to the nucleus.¹⁸ The faster migrating lower band was identified as ERCC1. As a negative control, the same region of the gel from the D-10 immunoprecipitate yielded no CCTα peptides, but ERCC1 was identified. This result suggests that 8F1, but not D-10, reacts with two distinct immune targets, ERCC1 and CCTα.

Figure 1. CCTα is the second antigen recognized by 8F1.

A: Whole cell lysate from A2780 cell line was subjected to immunoprecipitation with 8F1 or D-10, a specific anti-ERCC1 antibody. Precipitated proteins were resolved by electrophoresis, and visualized by gel silver staining. Note that 8F1 precipitated two proteins (Band 1 and Band 2) migrating at a molecular size near 37 kD, the size of ERCC1. The specific antibody D-10 precipitated only the lower band.

B: Results of mass spectrometry analysis for Band 1 and Band 2 are reported. For 8F1 (left column), the top band was mostly composed of CCTα, while the bottom was composed by a combination of both CCTα and ERCC1. For D-10 (right column) the majority of the lower band consisted of ERCC1.

C: Western blot of GFP tagged CCTα overexpression in HeLa cells. 8F1 (left) readily recognizes the transgene (CCTα-GFP). Note that only trace amounts of endogenous CCTα is expressed in this cell line.

D: Expression of CCTα by immunoblot in nine knock-down clones (CCTα shRNA clones) and one control (A2780). Note that as CCTα expression decreases on the blot probed with 8F1 (middle blot), the Band 2 disappears while Band 1 remains, showing that the CCTα knock-down is specific and ERCC1 levels remain constant (top blot). Tubulin is used as loading control (bottom blot).

E: Representative immunofluorescence images of CCTα knock-down clones (18-3 and 20-2) and control (A2780) with 8F1 and antibodies specific for CCTα and ERCC1 (FL297). Note that CCTα knock-down abrogate 8F1 signal (star), but ERCC1 level remains unchanged (arrow head).

We confirmed that 8F1 reacted with CCTα by stable expression of GFP-tagged CCTα in HeLa cells, which express only trace amounts of endogenous CCTα. As expected, recombinant CCTα, which migrates more slowly than the endogenous protein due to the GFP-tag, was readily detected by 8F1 (Fig. 1C, left panel). The identity was confirmed with a commercially available antibody recognizing CCTα (Fig 1C, right panel).

To determine the degree to which CCTα contributes to 8F1s nuclear signal, we knocked-down CCTα in A2780 cells using shRNA, which led to disappearance of the band corresponding to CCTα, while ERCC1 levels remained unchanged (Fig 1D). Individual stable clones, with varying levels of CCTα expression were then evaluated by immunofluorescence using 8F1 and the anti-CCTα antibody. The 8F1 nuclear signal was dramatically reduced in clones with reduced CCTα expression (Fig 1E), while ERCC1 expression assessed by the ERCC1-specific antibody FL297 remained unchanged (Fig 1E). We conclude that the in vitro 8F1 nuclear signal varies as a function of CCTα expression, even when ERCC1 expression remains constant. This demonstrates that 8F1 has two distinct immune targets, ERCC1 and CCTα. Both localize to the nucleus, interfering with the measurement of ERCC1 levels when 8F1 is used as the detection reagent.

8F1 Signal Intensity Depends on ERCC1 and CCTα Expression Levels in NSCLC

We next asked whether our in vitro findings are relevant for quantitation of ERCC1 levels by immunohistochemistry (IHC). Using knock-down cell lines, we characterized commercial antibodies raised against CCTα or ERCC1 to determine their specificity and suitability for IHC. Only background nuclear staining was observed in cells in which the respective protein had been substantially knocked down, while signal was readily visible in the control cells (Fig 2A and B, respectively). By IHC with an ERCC1-specific antibody (EP2143Y), the nuclear signal was unchanged in CCTα knock-down cells (Fig 2A), but dramatically decreased in ERCC1-deficient cells (Fig 2B). In contrast, when 8F1 was used, the nuclear signal decreased in CCTα knockdown cells (Fig 2A). These results show that the antibodies were specific for their respective antigens, that reduced CCTα levels do not affect ERCC1 expression, and that the 8F1 signal in IHC is a result of detecting both ERCC1 and CCTα.

Figure 2. Anti-ERCC1 and -CCTα antibodies are specific.

Specificity of anti-CCTα and anti-ERCC1 antibody (EP2134Y) by IHC in (A) CCTα knock-down or (B) ERCC1 knock out cells.

A: A2780 (WT; left column) and clone 20-2 (CCTα shRNA; right column) cells pellets were processed for IHC with the indicated antibodies. The CCTα antibody is specific as nuclear signal is lost in knock-down cells (stars). Note that while 8F1 signal is decreased, ERCC1 level (EP2143Y) remained unchanged (brown signal). Hematoxylin is used as counterstain (blue).

B: Specificity of anti-ERCC1 monoclonal antibody EP2143Y. Normal cells (ERCC1 WT) and XP2YO ERCC1 knock out (ERCC1-deficient from XP-F patient) skin fibroblast pellets were processed for IHC with EP2143Y antibody. Note that the brown horseradish peroxidase nuclear signal is reduced to background in the knockout cells while obvious in the WT cells.

C: CCTα staining of a squamous cell carcinoma of the lung with an AQUA score of 15,299 (spot #206; blue, nuclei; green, cytokeratin; red, CCTα).

D: CCTα staining of an adenocarcinoma of the lung with an AQUA score of 6,379 (spot #140; blue, nuclei; green, cytokeratin; red, CCTα).

Since 8F1 has been used extensively for determination of ERCC1 levels in NSCLC, we examined the extent to which 8F1 signal intensity estimated ERCC1 protein expression by IHC. We used a well-characterized cohort of 187 early stage NSCLC patients treated by surgery alone that was previously used to demonstrate a significant correlation between high 8F1 signal and improved patient survival.³ AQUA was performed on samples stained with specific anti-ERCC1 antibodies (FL297 or EP2143Y) or 8F1, and signal intensities were compared. There was a moderate, positive correlation between the signal intensities of the two ERCC1-specific antibodies (EP2143Y and FL297; rho = 0.44, p < 0.001). In contrast, the 8F1 signal had only a negligible to weak, positive correlation with the signal of either EP2143Y (rho = 0.19, p = 0.014) or FL297 (rho = 0.23, p = 0.002). However, there was a moderate, positive correlation between 8F1 and the CCTα signal (rho = 0.38; p < 0.001). This result supports the conclusion that the 8F1 nuclear signal is strongly influenced by the newly identified 8F1 antigen CCTα in early stage NSCLC.

CCTα Expression but not ERCC1 Depends on Tumor Histology

The NSCLC cohort comprises the histological subtypes squamous cell carcinoma, large cell carcinoma, and adenocarcinoma, the tumor stages IA and IB, and various other patient characteristics (Table 1). Tests for differences in expression revealed that CCTα levels were higher in squamous (Fig 2C) than in adeno- (Fig 2D) or large cell carcinomas (p ≤ 0.001; Fig 3). Tumor stage, sex, and other demographic and clinical variables were not associated with CCTα levels. In contrast, ERCC1 levels, as determined by EP2143Y or FL297 signal intensity, were independent of tumor subtype (Fig 3). CCTα significantly contributed to the 8F1 signal in squamous cell carcinomas, but not in adeno or large cell carcinomas.

Figure 3. 8F1, CCTα, EP2143Y, and FL297 levels by histology in NSCLC.

Dot blots showing the distribution of signal levels obtained with 8F1, CCTα, EP2143Y, and FL297 in 187 NSCLCs. Each dot represents the average of three replicates as determined by AQUA.

In HNSCC, the 8F1 Signal is Primarily a Result of CCTα

We analyzed the association between 8F1 signals and ERCC1 (EP2143Y) or CCTα expression levels in a cohort of 60 HNSCCs (Table 1).¹⁶ CCTα was very strongly and positively correlated with 8F1 (rho = 0.74, p ≤ 0.001), while the correlation between 8F1 and EP2143Y (ERCC1) was less strong (rho = 0.36, p = 0.011). These findings support the notion that CCTα expression is a major contributor to 8F1 immunostaining in both HNSCC and NSCLC. There was no significant association between 8F1, EP2143Y, or CCTα levels and patients’ sex or whether tumors were primary or recurrent. However, values for all three biomarkers were significantly lower in stage IV compared to stage I-III disease (Wilcoxon rank sum test p-values ≤0.001, 0.03, 0.007 for 8F1, EP2143Y or CCTα, respectively).

Correlation Between CCTα Expression and Clinical Outcomes in NSCLC and HNSCC

Based on numerous prior clinical studies using 8F1,^{1, 6} the influence of CCTα on 8F1 staining leads to the exciting possibility that CCTα expression levels may be of clinical value. To address this, we evaluated the association between CCTα or ERCC1 (EP2143Y) protein levels and clinical outcomes in a cohort in which 8F1 was previously reported to be prognostic of survival.

To test the association between CCTα or ERCC1 and DFS, we determined optimal cutoff levels in a multivariate Cox model that included age, sex, stage, and histology (for NSCLC only) or primary versus recurrence (for HNSCC only; Table 2). We were unable to establish an optimal cut point for EP2143Y indicating that ERCC1, as detected by EPY2143Y, is not suitable as a prognostic biomarker. However, CCTα was prognostic of DFS with an optimal cut-off level near the median AQUA value (1,388). The hazard ration (HR) for recurrence or death was 0.41 (95% CI: 0.23 – 0.73) for patients with high compared to low CCTα levels (Cox adjusted p = 0.002). Kaplan-Meier survival estimates showed a median DFS of 88.2 months (95% CI: 77.9 – not reached) for patients with high levels and 54.5 months (95% CI: 36.4 – 74.1) for those with low levels (Fig 4A; log-rank p = 0.002). The OS results were similar (Fig 4B) with a HR for death of 0.59 (95% CI 0.37 – 0.93; Cox adjusted p = 0.023) for patients with high compared to low CCTα levels (log-rank p = 0.056).

Table 2.

Multivariate Cox Model for DFS in NSCLC and HNSCC Patients

Variable	HR (95% CI)	p-Value
NSCLC
CCTα high vs low	0.41 (0.23 – 0.73)	0.002
Stage IB vs IA	1.93 (1.10 – 3.39)	0.022
Age > vs < median	1.00 (0.97 – 1.03)	0.916
Women vs Men	1.21 (0.68 – 2.14)	0.515
Large Cell vs Adeno	1.78 (0.87 – 3.62)	0.112
Squamous Cell vs Adeno	0.67 (0.35 – 1.29)	0.227
HNSCC
CCTα high vs low	0.37 (0.17 – 0.80)	0.012
Stage IV vs I-III	1.08 (0.52 – 2.24)	0.829
Age > vs < median	1.01 (0.98 – 1.04)	0.605
Women vs Men	1.05 (0.53 – 2.09)	0.891
Recurrent vs Primary	2.76 (1.00 – 7.58)	0.049

Figure 4. Kaplan-Meier survival estimates by high and low CCTα expression categories.

A: The median DFS for NSCLC patients with high versus low CCTα levels was 88.2 (95% CI: 77.9 – not reached) versus 54.5 (95% CI: 36.4 – 74.1) months.

B: The median OS for NSCLC patients with high versus low CCTα levels was 88.2 (95% CI: 77.9 – not reached) versus 69.4 (95% CI: 53.6 – 100.5) months.

C: The median DFS for HNSCC patients with high versus low CCTα levels was 19.0 (95% CI: 6.0 – not reached) versus 8.0 (95% CI: 3.4 – 14.0) months (log-rank p = 0.022, HR = 0.37, Cox adjusted p = 0.012).

D: The median OS for HNSCC patients with high versus low CCTα levels was 106.0 (95% CI: 88.0 – not reached) versus 17.1 (95% CI: 14.1 – 64.0) months (log-rank p = 0.006, HR = 0.41, Cox adjusted p = 0.027).

Similar results were obtained in HNSCC (Table 2, Fig 4 C-D) with an optimal CCTα IHC score cut-off value of 102, which is the 66^th percentile of CCTα levels.

DISCUSSION

DNA repair proteins, in particular ERCC1-XPF, are candidate predictive and prognostic biomarkers for management of patients with a variety of malignancies. ^{1-4, 6, 7, 15, 19, 20} Evidence from knockout animals and patients with genetic deficiency demonstrate that reduction of ERCC1-XPF expression is associated with susceptibility to cancer²¹ and hypersensitivity to radiation and cisplatin agents.²² This may explain the clinical behavior of NSCLC patients with low levels of tumoral ERCC1. These patients have poor prognosis when treated with surgery alone^{2, 3} and high response rates when treated with genotoxic agents.²

In the initial report using the NSCLC cohort,³ ERCC1 expression was measured using 8F1. The present report analyzed the same cohort using the ERCC1-specific antibodies EP2143Y and FL297, and demonstrates that ERCC1 expression is not prognostic of survival in NSCLC treated with surgery alone. This may be explained by the recent description of multiple ERCC1 isoforms, with only one contributing to DNA damage repair, and the apparent lack of isoform specificity of currently available commercial ERCC1 antibodies.⁸ Since our analysis was performed on tissue microarrays containing three replicates, we were able to compare the different antibodies in serial sections and found the correlation between specific ERCC1 antibodies to be strong and positive (rho = 0.44, p < 0.001) allowing us to determine that 8F1 is an inaccurate reagent for ERCC1 expression evaluation.

In HNSCC, two studies used specific antibodies to ERCC1¹⁴ or XPF¹⁵ and showed a correlation between ERCC1-XPF expression and clinical outcomes. In contrast, we find that 8F1 staining intensity does not correlate with survival. We also found no association between outcomes and ERCC1 levels using the specific antibody FL297, which is most likely a result of the above mentioned lack of isoform specificity.⁸ However, we must caution that our HNSCC cohort was not designed or powered to address this question, and patients were inconsistently treated with DNA damaging agents.

CCTα is the second nuclear antigen recognized by 8F1, which is consistent with a recent report using protein microarray technology.²³ CCTα is the major isoform of a rate-limiting enzyme in the main pathway for biosynthesis of phosphatidyl choline.²⁴ It participates in formation of the nuclear membrane phospholipid bilayer²⁵ and nucleoplasmic reticulum.²⁶ It is essential for embryo survival²⁷ and a key regulator of surfactant production by type II alveolar cells.²⁸ CCTα activity increases with H-Ras oncogene expression. It is also involved in anchorage-independent growth of Ras-transformed cells,²⁹ in resistance to apoptosis,³⁰ in cell proliferation,³¹ and in synthesis of phosphatidyl choline, a key substrate in signaling cascades involved in cancer.³²

We found that CCTα has higher expression in squamous cell carcinoma than adenocarcinoma, which is consistent with publicly reported mRNA expression data (www.ncbi.nlm.nih.gov/geo/tools/profileGraph.cgi?ID=GDS3627:204210_s_at).³³ Since survival among the histologic subtypes in our NSCLC cohort was not different, we conclude that the association between CCTα levels and survival is unlikely to be due to its ability to discriminate tumors based on histology.

We demonstrate that CCTα has a protective effect on patient outcomes, with high expressing patients having longer DFS and OS using an optimal cut-point analysis and a multivariate Cox model taking other clinical variables often associated with patients’ survival into account. This is even more remarkable considering that the studies on NSCLC and HNSCC were conducted at two separate institutions, using different IHC and quantitation methods. Our findings should be confirmed in large, prospective cohorts. This is the first report that CCTα is a biomarker in solid tumors and could inform treatment decisions by identifying patients with NSCLC, HNSCC, and other solid tumors at risk for recurrence.

In summary, we identified the second antigen recognized by the monoclonal antibody 8F1 as CCTα, and demonstrated that CCTα expression introduces a signal interfering with ERCC1 quantitation by 8F1. We confirmed that 8F1 is a poor estimator of ERCC1 levels in clinical samples. We provide provocative data indicating that CCTα may be a more important factor in determining clinical outcomes in NSCLC and HNSCC than ERCC1. We believe that CCTα has value as a biomarker and deserves careful further evaluation in large cohorts of NSCLC and other solid tumors, especially tumors treated with genotoxic agents.