Low ERCC1 mRNA and protein expression are associated with worse survival in cervical cancer patients treated with radiation alone

Radiotherapy and Oncology 97 (2010) 352–359

Contents lists available at ScienceDirect

Radiotherapy and Oncology journal homepage: www.thegreenjournal.com

Clinical radiobiology

Low ERCC1 mRNA and protein expression are associated with worse survival in cervical cancer patients treated with radiation alone Corinne M. Doll a,⇑, Michael Prystajecky b, Misha Eliasziw a, Alexander C. Klimowicz b, Stephanie K. Petrillo b, Peter S. Craighead a, Desiree Hao a, Roman Diaz b, Susan P. Lees-Miller c,d, Anthony M. Magliocco a,b a

Department of Oncology; b Department of Pathology; c Department of Biochemistry and Molecular Biology; and d Department of Oncology, University of Calgary, Alberta, Canada

a r t i c l e

i n f o

Article history: Received 24 June 2010 Received in revised form 23 August 2010 Accepted 24 August 2010 Available online 9 October 2010 Keywords: Cervical cancer ERCC1 Radiation therapy

a b s t r a c t Purpose: To evaluate the association of excision repair cross-complementation group 1 (ERCC1) expression, using both mRNA and protein expression analysis, with clinical outcome in cervical cancer patients treated with radical radiation therapy (RT). Experimental design: Patients (n = 186) with locally advanced cervical cancer, treated with radical RT alone from a single institution were evaluated. Pre-treatment FFPE biopsy specimens were retrieved from 112 patients. ERCC1 mRNA level was determined by real-time PCR, and ERCC1 protein expression (FL297, 8F1) was measured using quantitative immunohistochemistry (AQUAÒ). The association of ERCC1 status with local response, 10-year disease-free (DFS) and overall survival (OS) was analyzed. Results: ERCC1 protein expression levels using both FL297 and 8F1 antibodies were determined for 112 patients; mRNA analysis was additionally performed in 32 patients. Clinical and outcome factors were comparable between the training and validation sets. Low ERCC1 mRNA expression status was associated with worse OS (17.9% vs 50.1%, p = 0.046). ERCC1 protein expression using the FL297 antibody, but not the 8F1 antibody, was significantly associated with both OS (p = 0.002) and DFS (p = 0.010). After adjusting for pre-treatment hemoglobin in a multivariate analysis, ERCC1 FL297 expression status remained statistically significant for OS [HR 1.9 (1.1–3.3), p = 0.031]. Conclusions: Pre-treatment tumoral ERCC1 mRNA and protein expression, using the FL297 antibody, are predictive factors for survival in cervical cancer patients treated with RT, with ERCC1 FL297 expression independently associated with survival. These results identify a subset of patients who may derive the greatest benefit from the addition of cisplatin chemotherapy. Ó 2010 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 97 (2010) 352–359

Cervical cancer is the third most common cancer in women worldwide and accounts for over 190,000 deaths annually [1]. A decade ago, clinical trials demonstrated that the addition of concomitant chemotherapy to RT regimens improves both progression-free and overall survival [2,3]. However, these trials also found increased acute toxicity in patients treated with concomitant chemoradiation therapy (CRT) as compared to those treated with RT; moreover, survival rates have not improved over the last decade. Predictive tumor markers of radiation responsiveness would be of significant clinical value, as they would allow patients who are likely to respond to RT alone to be identified, thereby avoiding toxicity of additional therapies. This would enable better selection of patients who are most likely to benefit most from the

⇑ Corresponding author. Address: Department of Oncology, Tom Baker Cancer Centre, Room CC113A, 1331 29th Street NW, Calgary, Alberta, Canada T2N 4N2. E-mail address: [email protected] (C.M. Doll). 0167-8140/$ - see front matter Ó 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2010.08.019

addition of chemotherapy, leading to improved treatment response. Excision repair cross-complementation group 1 (ERCC1) is a protein that plays a role in several DNA repair pathways. ERCC1 associates with the protein xeroderma pigmentosum complementation group F (XPF) to form a structure-specific heterodimeric endonuclease that cleaves damaged DNA 50 to the lesion [4]. XPF possesses endonuclease catalytic activity [5], whereas ERCC1 is responsible for DNA binding and stabilizing XPF [6,7]. In nucleotide excision repair, the ERCC1-XPF complex plays a critical role in the removal of DNA intrastrand crosslinks [8]. The ERCC1-XPF endonuclease has also been implicated in homologous recombination [9,10], interstrand crosslink repair [11], and more recently, in the repair of DNA double-strand breaks via a Ku-independent endjoining pathway [12]. Several retrospective studies have investigated the prognostic and predictive values of ERCC1 mRNA and protein expression in response to cancer therapies. High ERCC1 expression has been

C.M. Doll et al. / Radiotherapy and Oncology 97 (2010) 352–359

shown to be associated with poor treatment response and survival in colorectal [13], gastric [14,15], ovarian [16], esophageal [17], head and neck [18,19], and lung cancer [20–23] patients treated with platinum-based chemotherapy. Although the predictive value of ERCC1 expression has not been investigated in patients with locally advanced cervical cancer, there is evidence that human cervical tumor cells with high ERCC1 mRNA are resistant to platinum-based chemotherapy [24]. The relationship between high ERCC1 expression and cisplatin resistance can be attributed to enhanced removal of platinum-induced DNA adducts by the nucleotide excision repair pathway [8]. Radiation therapy kills tumor cells by inducing several types of DNA damage, of which double-strand breaks are believed to be the most lethal [25]. Given the role of ERCC1 in the repair of these lesions [12], it is possible that ERCC1 status may be associated with RT response, although this relationship is likely complex. Mammalian cell lines deficient in ERCC1 demonstrate an increased sensitivity to ionizing radiation under hypoxic conditions and fail to exhibit an adaptive response to radiation [26–28]. In a murine xenograft model, upregulated ERCC1 expression is associated with radioresistance in tumors derived from cervical carcinoma cells [29]. A study of head-and-neck cancer patients treated with radiation therapy alone found that polymorphisms in ERCC1 are significantly associated with better progression-free and overall survival [30]. However, the predictive and prognostic values of ERCC1 expression in cervical cancer patients treated with RT alone have not been previously investigated. The most commonly used antibody against ERCC1 in IHC studies, 8F1, has been recently shown to stain in a non-specific fashion [31,32]. These studies have demonstrated that 8F1 detects not only ERCC1, but also a second major antigen of unknown identity. The anti-ERCC1 antibody clone FL297 appears to have a greater specificity to ERCC1 in IHC applications [32]. In the present study, we evaluated the association of ERCC1 mRNA with protein expression in cervical cancer patients treated with radical RT alone. In addition, using AQUAÒ technology, we quantitatively analyzed the expression of ERCC1 using both 8F1 and FL297 antibodies. Finally, the clinical significance of both ERCC1 mRNA and protein expression was determined. Materials and methods Patients, treatment, and biosamples One hundred and eighty-six patients who completed radical RT alone, with curative intent, for cervical cancer between 1986 and 1996 at a single institution (Tom Baker Cancer Centre, Calgary, AB) were identified from the Alberta Cancer Registry database. None of the patients were treated with primary surgery or with concurrent chemotherapy. Patients were excluded if they had prior definitive treatment for cervical cancer, active malignancy at another site, and/or less than five years of follow-up data. A chart review was performed, and clinical, treatment, and outcome data were recorded. Standard staging procedures were examination under anaesthesia, cystoscopy, sigmoidoscopy, chest X-ray, and CT abdomen and pelvis. Full scientific and ethics approval for this study was granted by the University of Calgary Office of Medical Bioethics. Radiotherapy was given as a combination of external beam RT (EBRT) and low-dose rate (LDR) brachytherapy using the cesium SelectronÒ (Nucletron, The Netherlands) unit. The most common external beam RT dose and fractionation schedule was 45 Gy in 25 fractions over 5 weeks. The most common LDR brachytherapy dose and fractionation was 40 Gy in 2 fractions, 1 week apart, starting 1 week after the completion of external beam RT. Mean total RT dose to Point A was 82 Gy. After the completion of RT, patients were followed with clinical examination every three

353

months for the first year, q4 monthly for year 2, then q6 monthly to 10 years. Follow-up imaging was performed as clinically indicated. Response to RT was determined clinically at 3 months after the completion of radiotherapy, by a radiation oncologist. Pre-treatment formalin-fixed paraffin-embedded (FFPE) tumor biopsy specimens were retrieved and reviewed. At least three 0.6 mm representative cores were taken from each available block containing tumor areas and placed in a tissue microarray (TMA) using a Beecher Manual Tissue Microarrayer (Beecher Instruments, Inc.). For mRNA analysis, 3–6  10–20 lm whole sections were taken from the tumor blocks. RNA extraction and cDNA synthesis Total RNA was isolated from FFPE tissue specimens using the RecoverAll™ kit (Ambion, Austin, TX) according to the manufacturer’s instructions. Isolated RNA was subsequently quantified by spectrophotometry (Biophotometer; Eppendorf, Mississauga, Ontario, Canada). Samples with concentrations of 100 ng/ll or higher and A260/280 nm ratios greater than 1.7 were analyzed for ERCC1 mRNA expression. In preparation for reverse transcription, isolated RNA was diluted to 100 ng/ll with nuclease-free water. Reverse transcription was performed with the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA) using random primers and 1 lg of RNA. Real-time PCR To identify a suitable gene for normalizing mRNA expression data, the gene expression levels of 11 commonly used ‘‘housekeeping” genes were evaluated in 8 randomly selected cervical tumor specimens using the TaqMan Human Endogenous Control Plate (Applied Biosystems). The analyzed genes were 18S rRNA (18S), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), hypoxanthine ribosyl transferase (HPRT1), beta-glucuronidase (GUSB), beta-actin (ACTB), beta-2-microglobulin (B2M), phosphoglycerokinase (PGK1), large ribosomal protein (RPLP0), TATA-box binding protein (TBP), transferrin receptor (TFRC), and cyclophilin A (PPIA). Real-time PCR was performed using an ABI 7500 Sequence Detection System (Applied Biosystems) and plates were thermocycled as follows: 1 cycle of 2 min at 50 °C and 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C. Sequence Detection Software (Applied Biosystems) was used to calculate fluorescence threshold (CT) values. The mean CT values of duplicate reactions were transformed into non-normalized gene expression values using the comparative DCT method as described previously [33]. Gene expression stability analysis was then performed using the geNorm and NormFinder applications [34,35]. The expression levels of all candidate normalization genes, with the exception of PPIA and TBP, were determined for each of the 8 analyzed samples. The geNorm and NormFinder applications identified ACTB as the most stably expressed gene of the 9 genes that amplified in all of the analyzed samples (Supplementary Table 1). Samples were subdivided equally into two subgroups according to the ages of the samples for NormFinder analysis [36]. On the basis of the results of the geNorm and NormFinder analysis, ACTB was used to normalize ERCC1 gene expression measurements. Relative ERCC1 mRNA levels were measured using fluorescencebased, TaqMan assays (Applied Biosystems). On the basis of the results obtained from the TaqMan Endogenous Control Plate assay, ACTB was chosen as an endogenous control for gene expression analysis. TaqMan assays with intron spanning primers were selected for ACTB (4326315E) and ERCC1 (Hs01012158_m1) to avoid genomic DNA contamination. Reactions were prepared according to the manufacturer’s instructions and contained 100 ng of cDNA (assuming 100% efficiency during reverse transcription), 25 ll of

354

ERCC1 mRNA and protein expression in cervical cancer

TaqMan Gene Expression Mastermix, 2.5 ll of 20 primers and probe, and 20.5 ll of nuclease-free water (all Applied Biosystems). Thermocycling was carried out as described above. All reactions were performed in triplicate and a no template control was included in each run to detect contamination of reagents. ERCC1 mRNA levels were calculated by the comparative threshold cycle method using ACTB as the endogenous control and a single benign cervix specimen from a hysterectomy specimen as the calibrator sample [37]. Gene expression levels were reported as the ratio of test case to normal cervix. Gene expression measurements were considered unreliable when ACTB CT values were greater than 34, as higher CT values are indicative of mRNA transcript levels approaching single copy numbers. Fluorescent immunohistochemistry for ERCC1 Tissue microarray slides were deparaffinized in xylene, rinsed in ethanol, and rehydrated. Antigen retrieval was performed by heating slides to 121 °C in a Tris/EDTA-based (pH 9.0) Target Retrieval Solution (Dako, Glostrup, Denmark) for 3 min in a pressure cooker. Slides were then rinsed in water and loaded into a Dako Autostainer. Endogenous peroxidase activity was quenched using the peroxidase block from the DAKO EnVision TM + System (Dako) for 10 min and non-specific staining was further reduced with a 15 min incubation in Signal Stain protein block and antibody diluent (Cell Signaling, Danvers, MA). Slides were washed in Dako wash buffer and stained for 1 h at room temperature. For 8F1 staining, the 8F1 antibody (Neomarkers, Fremont, CA) was diluted 1:100 in Signal Stain with a 1:200 dilution of anti-pan-cytokeratin rabbit polyclonal antibody. For FL297 staining, the FL297 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was diluted 1:100 in Signal Stain with a 1:200 dilution of anti-pan-cytokeratin mouse monoclonal AE1/AE3 antibody (Dako). After washing in Dako wash buffer, corresponding secondary antibodies were applied for 1 h at room temperature. For 8F1, a goat anti-mouse antibody conjugated to a horseradish peroxidase-decorated dextran polymer backbone from the Dako EnVision TM + system (Dako) and a 1:200 dilution of Alexa-555 conjugated goat anti-rabbit antibody (Invitrogen, Carlsbad, CA) were used. For FL297, a goat anti-rabbit antibody conjugated to a horseradish peroxidase-decorated dextran polymer backbone from the Dako EnVision TM + system (Dako) and a 1:200 dilution of Alexa-555 conjugated goat anti-mouse antibody (Invitrogen) were used. After further rinsing in wash buffer (Dako), slides were incubated for 10 min with the TSA-Plus Cy5 tyramide signal amplification reagent (PerkinElmer, Waltham, MA). After three additional washes in wash buffer (Dako), the tissue microarray slides were mounted with coverslips using Vectashield anti-fade mounting medium containing DAPI (Vector Laboratories, Burlingame, CA). Automated image acquisition and analysis (AQUAÒ) AQUAÒ was performed using the HistoRx PM-2000™, which has previously been described in detail [38]. Briefly, high resolution monochromatic 8-bit (resulting in 256 discrete intensity values per pixel of an acquired image) digital images were obtained for every histospot on the tissue microarrays using filters specific for 40 ,6-diamidino-2-phenylindole (DAPI) to define the nuclear compartment, Alexa-555 to define pan-cytokeratin-positive cervical carcinoma cells and the tumor cytosolic compartment, and Cy5 to define the target biomarker ERCC1. Analysis of the digital images was performed with AQUAÒ script as previously described [38]. Briefly, a tumor-specific mask was generated to distinguish the carcinoma cells from supporting stromal tissue by thresholding the pan-cytokeratin images. Thresholding created a binary mask that identified the presence or absence of tumor cells by the

presence of a pixel that was ‘on’ or ‘off’, respectively. Thresholding levels were verified, and adjusted if necessary, by spot-checking a small sample of images to determine an optimal threshold value. All images were then processed using this optimal threshold value. Next, the in- and out-of-focus images taken of the compartmentspecific tags (pan-cytokeratin and DAPI) and the target marker were processed using an algorithm called RESA (Rapid Exponential Subtraction Algorithm) [38]. The PLACE (Pixel-based Local Assignment for Compartmentalization of Expression) algorithm then assigned each pixel in the target images to a specific subcellular compartment. Once each pixel was assigned to a subcellular compartment, the signal in each location was tabulated and used to generate compartment-specific AQUAÒ scores, which reflect the average signal intensity per compartment area. Images were validated according to the following: (1) >10% tumor area covered, (2) >40% of the image was usable (i.e., not compromised due to overlapping or out-of-focus tissue).

Statistical analysis As quantitative ERCC1 expression in cervical cancer is novel, there are no established cutpoints available. Although several different methods exist to determine cutpoints, namely biological determination, data-oriented, and outcome-oriented, there is no single method or criterion to specify which approach is the best. For the present analyses, we used a data-oriented approach to select the cutpoint, corresponding to the lower quartile (25th percentile) of the ERCC1 expression data. Associations between dichotomized ERCC1 mRNA, FL297, and 8F1 levels and clinicopathological variables were assessed for statistical significance using a chi-square test. Overall survival (OS) was measured from the date of diagnosis to the date of death or date of last follow-up. Diseasefree survival (DFS) was measured from the date of diagnosis to the date of relapse, progression, death, or last follow-up. Kaplan–Meier event-free survival curves were used to estimate 10-year rates and the logrank test was used to compare the curves for statistical significance. Cox proportional hazard regression was used to adjust the results for prognostic factors [39,40] that were identified as confounders. Response to RT was assessed by clinical examination at 3 months following the completion of treatment and defined as follows: a complete response was the absence of clinical evidence of residual tumor; partial response was a reduction in the size of the original tumor by >50% (but less than 100%); and non-response was a <50% reduction in the size of the original tumor. Levels of agreement among dichotomized ERCC1 mRNA, FL297, and 8F1 scores were assessed using a kappa (j) statistic. Commonly used benchmarks for interpreting j agreement are: 0.0–0.20 slight, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 substantial, and 0.81–1.00 almost perfect. Statistical analyses were performed using SAS 9.2 software (SAS Institute Inc., Cary, NC, USA.). All reported p-values are two-sided and p < 0.05 were considered statistically significant. Results Of the 186 patients identified from the database, 112 had pretreatment FFPE tumor biopsy specimens available for the analysis for evaluation of ERCC1 by AQUAÒ technology, using both FL297 and 8F1 antibodies. Of these 112 FFPE biopsy specimens, 63 (56%) were of sufficient size (at least 1 cm) and tumor density (at least 50% tumor) to be additionally selected for mRNA expression analysis. ERCC1 mRNA expression levels were successfully measured from 32 of these specimens. ERCC1 mRNA levels could not be determined for the remaining specimens due to either an insufficient yield of RNA or a failure to detect the endogenous control

C.M. Doll et al. / Radiotherapy and Oncology 97 (2010) 352–359

(ACTB) at sufficient levels. For the purpose of analyses, the sample of 32 patients who had each of mRNA, FL297, and 8F1 measured were considered as the training set, and the remaining sample of 80 patients who did not have mRNA measured were considered as the validation set. The staining patterns for both 8F1 and FL297 were predominantly nuclear with a faint cytosolic signal. Subtle differences in the intensity of the nuclear signal were observed on the same patients’ tissue cores stained with the two antibodies (Fig. 1A). While FL297 staining was always predominantly nuclear, 8F1 demonstrated clear examples of cytosolic staining in the absence of strong nuclear staining (Fig. 1B). The baseline clinical and tumor characteristics are shown in Table 1. The median scores for ERCC1 mRNA, FL297, and 8F1 are shown in Table 1, and the corresponding lower quartiles are reported in the footnote of Table 2. There were no apparent differences between patient and tumor characteristics in the training set and the validation set. The overall median follow-up of patients was 5.2 years (4.8 years in the training set and 5.3 in the validation set). The 10-year OS rates for the training and validation sets were 44.1% and 43.0%, respectively (p = 0.74); median survivals were 5.6 years and 7.0 years, respectively. Additionally, the OS rates, when stratified by the lower quartile cutpoint for FL297 and 8F1, were similar

355

between the training and validation sets (Table 2). The similar results between the training and validation sets imply that the data in the training set are representative and comparable to the larger dataset. Therefore, the remaining analyses were performed on the 32 patients who had mRNA evaluated, and the aggregate of 112 patients who had ERCC1 protein expression analysis using both FL297 and 8F1 antibodies. Among the 112 patients, 95 (84.4%) had complete local response. Complete response rates for patients with low vs high ERCC1 mRNA status was 75.0% vs 87.5%, p = 0.40. There was no difference in the complete response rates by FL297 expression status (85.7% vs 84.5%, p = 0.88) or by 8F1 expression status (78.6% vs 86.9% p = 0.29). Patients with low ERCC1 mRNA expression levels had significantly worse OS (17.9% versus 50.1%, p = 0.046, Table 3 and Fig. 2A) and worse DFS (21.4% versus 47.4%, p = 0.083, Table 3 and Fig. 2B) than those with higher expression levels. Similarly, patients with low FL297 scores had significantly worse OS (p = 0.002) and worse DFS (p = 0.010) than those with higher FL297 scores (Table 2 and Fig. 2C and D). In contrast, 8F1 expression status was not associated with either OS (47.6% versus 42.7%, p = 0.94, Table 3 and Fig. 2E) or DFS (44.4% versus 40.0%, p = 0.98, Table 3 and Fig. 2F).

Fig. 1. Comparison of 8F1 and FL297 staining on the same histospots. (A) An example of FL297 and 8F1 demonstrating a similar nuclear staining pattern. (B) An example of differential staining between 8F1 and FL297; 8F1 demonstrating a distinct cytosolic staining pattern compared with the nuclear staining pattern of FL297. The Merge panels are pseudo-colored blue for DAPI, green for pan-cytokeratin and red for 8F1 or FL297 staining. Contrast enhancement was applied to the DAPI images for clarity.

356

ERCC1 mRNA and protein expression in cervical cancer

Table 1 Baseline patient and tumor characteristics in training and validation sets.

Table 3 Univariate results from Kaplan–Meier analysis for overall survival (OS) and diseasefree survival (DFS) using entire set.

All patients (N = 112) n (%)

Training set (N = 32) n (%)

Validation set (N = 80) n (%)

pValue*

Age (years) 650 >50

47 (42.0) 65 (58.0)

13 (40.6) 19 (59.4)

34 (42.5) 46 (57.5)

0.86

Histology Squamous Adeno

99 (88.4) 13 (11.6)

27 (84.4) 5 (15.6)

72 (90.0) 8 (10.0)

0.40

Stage IB–IIA IIB–III

41 (36.6) 71 (63.4)

12 (37.5) 20 (62.5)

29 (36.3) 51 (63.7)

0.90

Tumor size (cm) 64.0 >4.0

57 (50.9) 55 (49.1)

16 (50.0) 16 (50.0)

41 (51.2) 39 (48.8)

0.90

Nodal involvement Absent Present

99 (88.4) 13 (11.6)

29 (90.6) 3 (9.4)

70 (87.5) 10 (12.5)

0.64

8F1 AQUA (N = 112)c 6Lower 28 (25.0) quartile >Lower 84 (75.0) quartile

Pre-treatment Hb (g/L) 6115 27 (24.1) >115 85 (75.9)

0.73

44.7 42.9

45.3 38.9

0.86

20 (25.0) 60 (75.0)

47 (42.0) 65 (58.0)

0.91

7 (21.9) 25 (78.1)

Age (years) 650 >50 Histology Squamous Adeno

99 (88.4) 13 (11.6)

45.9 28.8

0.77

43.0 27.7

0.60

Stage IB–IIA IIB–III

41 (36.6) 71 (63.4)

52.9 37.1

0.097

53.3 33.9

0.059

Tumor size (cm) 64.0 57 (50.9) >4.0 55 (49.1)

54.5 31.8

0.008

52.5 28.7

0.002

Nodal involvement Absent 99 (88.4) Present 13 (11.6)

44.1 35.2

0.39

42.0 30.8

0.30

Pre-treatment Hb (g/L) 6115 27 (24.1) >115 85 (75.9)

10.6 53.5

<0.001

10.3 52.0

<0.001

ERCC1 mRNA, median (IQR)

5.13 (7.62)

5.13 (7.62)

–

–

ERCC1 FL297 AQUA, median (IQR)

38.15 (36.39)

33.49 (32.26)

38.87 (39.16)

0.59

ERCC1 8F1 AQUA, median (IQR)

779.11 (382.01)

757.99 (316.23)

797.55 (390.48)

0.88

Number N (%) ERCC1 mRNA (N = 32)a 6Lower 8 (25.0) quartile >Lower 24 (75.0) quartile FL297 AQUA (N = 112)b 6Lower 28 (25.0) quartile >Lower 84 (75.0) quartile

Squamous = squamous cell carcinoma. Adeno = adenocarcinoma and adenosquamous. Hb = hemoglobin. IQR = interquartile range. * p-Value comparing training set to validation set.

Table 2 Overall survival (OS) in training and validation sets.

a b

Training set (N = 32) Number N (%) ERCC1 mRNAa 6Lower 8 (25.0) quartile >Lower 24 (75.0) quartile FL297 AQUAb 6Lower 8 (25.0) quartile >Lower 24 (75.0) quartile 8F1 AQUAc 6Lower quartile >Lower quartile a b c

Validation set (N = 80)

c

10-Year OS (%)

pValue

Number N (%)

10-Year OS (%)

pValue

17.9

0.046

–

–

–

–

–

20 (25.0)

26.7

60 (75.0)

51.1

50.1

28.6

0.45

51.5

0.006

10-Year OS (%)

pValue

10-Year DFS (%)

pValue

17.9

0.046

21.4

0.083

50.1

15.3

47.4

0.002

52.0

47.6

23.1

0.010

48.4

0.94

42.7

44.4

0.98

40.0

Lower quartile = 3.06. Lower quartile = 22.91. Lower quartile = 609.48.

Lower expression of ERCC1 mRNA and FL297 approximately doubled the hazard of mortality and relapse. Agreement between dichotomized ERCC1 mRNA expression and FL297 scores was moderate (j = 0.45, p-value = 0.009). There was no agreement between ERCC1 mRNA and 8F1 (j = 0.00, p-value = 1.00) and FL297 and 8F1 (j = 0.00, p-value = 0.31).

Discussion

8 (25.0)

37.5

24 (75.0)

47.5

0.33

20 (25.0)

51.4

60 (75.0)

44.2

0.67

Lower quartile training set = 3.06. Lower quartile training set = 25.67, validation set = 21.39. Lower quartile training set = 609.48, validation set = 608.37.

Both tumor size and pre-treatment hemoglobin level emerged as significant prognostic factors for both overall survival and disease-free survival in the univariate analysis (Table 3). As tumor size was not a confounding factor, only pre-treatment hemoglobin was included in the multivariate analyses. Results from Cox proportional hazards regression showed that both ERCC1 mRNA and FL297 remained strong prognostic factors even after adjusting for pre-treatment hemoglobin levels, whereas 8F1 did not (Table 4).

In the present study, we found that low ERCC1 mRNA and protein expression are associated with worse survival in cervical cancers treated with radiation therapy, with ERCC1 expression measured using the FL297 antibody independently associated with OS on multivariate analysis. This is the largest study of ERCC1 in cervical cancer where the expression was comprehensively measured using multiple assay methods. At first glance, these findings seem surprising given the pre-clinical evidence implicating ERCC1 overexpression in radioresistance [26–29,41]. However, an association between high ERCC1 expression and improved survival has also been reported in non-small lung cancers treated with surgical resection or combined platinum- and taxane-based therapy [23,42– 44]. High ERCC1 expression has also been correlated with favorable prognostic factors in breast cancer patients treated with adjuvant RT [45]. One proposed mechanism for the relationship between high ERCC1 expression and better outcome in these cancers is that

C.M. Doll et al. / Radiotherapy and Oncology 97 (2010) 352–359

a loss of DNA-repair capacity leads to the accumulation of genetic damage, which in turn may result in the emergence of an aggressive tumor phenotype [42]. This hypothesis is supported by several lines of evidence. First, ERCC1- deficient stem cells have an elevated frequency of spontaneous chromosome aberrations as compared to wild-type cells [46]. A separate study demonstrated that Chinese hamster ovary cells deficient for ERCC1 had a higher rate of mutagenesis than the wild-type cells [28]. Finally, a case control study conducted by Cheng et al. [47] revealed a trend towards reduced ERCC1 mRNA expression in the peripheral lymphocytes of newly

357

diagnosed lung cancer patients as compared to cancer-free controls. We, therefore, hypothesize that the poor outcomes observed for patients with low ERCC1 expression are not directly related to the repair of radiation-induced DNA damage by ERCC1-dependent DNA repair pathways, but a result of a more aggressive tumor phenotype that is an emergent property of reduced DNA-repair capacity. The interplay between low ERCC1 level in terms of genetic instability and positive therapeutic response with the addition of cisplatin to RT is not clear. Whereas high ERCC1 status has been shown to predict resistance to platinum-based chemotherapy in

Fig. 2. Kaplan–Meier overall survival and disease-free survival curves stratified by ERCC1 mRNA (A and B), stratified by ERCC1 FL297 AQUA (C and D), and stratified by ERCC1 8F1 AQUA (E and F).

358

ERCC1 mRNA and protein expression in cervical cancer

Table 4 Multivariate results from cox regression for overall survival (OS) and disease-free survival (DFS), adjusted for confounding factors, using entire set.

ERCC1 mRNAa 6Lower quartile >Lower quartile

OS Hazard ratio (95% CI)

pValue

DFS Hazard ratio (95% CI)

pValue

2.4 (0.8–7.6)

0.13

2.2 (0.7–6.7)

0.16

1.0

Pre-treatment Hb (g/L) 6115 3.4 (1.2–9.8) >115 1.0

1.0

0.023

3.3 (1.2–9.2) 1.0

0.024

0.031

1.5 (0.9–2.7)

0.13

FL297 AQUAb 6Lower quartile >Lower quartile

1.9 (1.1–3.3) 1.0

Pre-treatment Hb (g/L) 6115 3.5 (2.0–6.2) >115 1.0 8F1 AQUAc 6Lower quartile >Lower quartile

1.3 (0.7–2.5)

1.0

<0.001

4.3 (2.5–7.5) 1.0

<0.001

0.37

1.3 (0.7–2.5)

0.33

1.0

Pre-treatment Hb (g/L) 6115 4.2 (2.4–7.4) >115 1.0

1.0

<0.001

5.0 (2.9–8.7) 1.0

<0.001

Hb = hemoglobin. a Lower quartile = 3.06; univariate hazard ratio OS = 2.9, p-value = 0.056; DFS = 2.5, p-value = 0.094. b Lower quartile = 22.91; univariate hazard ratio OS = 2.3, p-value = 0.002; DFS = 2.0, p-value = 0.012. c Lower quartile = 609.48; univariate hazard ratio OS = 1.0, p-value = 0.94; DFS = 1.0, p-value = 0.98.

other tumor types, further data looking at this biomarker in cervical cancer patients treated with CRT are required to clarify this issue. Our results contribute to the growing body of evidence that suggests that the most commonly used ERCC1 antibody, 8F1, is unsuitable for measuring ERCC1 protein expression in IHC applications. The 8F1 antibody has previously been shown to detect at least two proteins with different molecular weights [31]. A recent study confirmed that 8F1 is not specific for ERCC1 and confirmed that the FL297 antibody is appropriate for detecting ERCC1 FFPE samples [32,32]. In the present study, we observed that FL297 AQUAÒ scores, but not 8F1 AQUAÒ scores, correlated with ERCC1 mRNA levels. We also observed that FL297 AQUAÒ scores and ERCC1 gene expression levels predict for disease-free and overall survival in a similar fashion. Thus, our results support the emerging data indicating that FL297, but not 8F1 is a valid antibody for detecting ERCC1 protein levels in IHC based assays. In our study ERCC1 expression was evaluated by both quantitative IHC and real-time PCR. Although ERCC1 mRNA and protein expression levels (as determined using the FL297 antibody) were correlated, we believe that quantitative IHC is a better methodology for evaluating ERCC1 expression for several reasons. First, ERCC1 protein expression more directly reflects cellular DNArepair capacity than mRNA expression levels. The use of IHC and TMAs also circumvents problems associated with stromal contamination of specimens, as the protein of interest can be measured directly within cancer cells. In contrast, RNA extracted for gene expression analysis typically requires whole sections of biopsy specimens, and as such have a higher degree of stromal contamination, diluting the measurement from the cancer cells. Lastly, IHC analysis requires less tissue than real-time PCR, thereby allowing

more samples from a given clinical tumor dataset to be analyzed, and is more amenable to FFPE samples [48]. While our results suggest a predictive role for ERCC1 mRNA and protein level in cervical cancer patients treated with RT alone, the limitations of the study should be recognized. First, the mRNA study analyzed only 32 of the 112 available biopsy specimens due to either inadequate tumor size or failure to obtain validated ERCC1 mRNA expression levels, the latter presumably arising from sample degradation. Again, this issue is a likely function of the stringency of our selection of specimens of sufficient quantity and >50% tumor on histological assessment. Nonetheless, we were still able to detect a significant difference in survival according to ERCC1 mRNA and protein expression. While the sample size of our study was somewhat limited, patients in this study were treated in a uniform manner at a single institution. These findings provide an insight into how tumor ERCC1 expression affects response to RT, and likely CRT, and suggest that ERCC1 expression may be a candidate-stratification factor for the treatment of cervical cancer. The feasibility of using ERCC1 mRNA expression assays to stratify treatment and improve treatment response has recently been proven in non-small-cell lung cancers treated with chemotherapy [49]. Further investigation is required to determine whether these assays are sufficiently reliable to use routinely as a basis to select specific patient treatments. In conclusion, this study shows that low tumoral ERCC1 expression, measured by either mRNA or IHC AQUAÒ using FL297 but not with the 8F1 antibody, is associated with worse survival in cervical cancer patients treated with RT. Patients with low ERCC1 expression have poor outcomes, and are most likely to derive the greatest benefit from concomitant CRT therapy, since they are less likely to be cisplatin resistant. Moreover, we identify a subset of patients with comparatively better outcome following treatment with radical RT alone, for whom the addition of cisplatin chemotherapy may not be as beneficial. Financial support Funded in part by the Alberta Cancer Foundation and the Southern Alberta Cancer Research Institute. Conflict of interest statement None of the authors (C.D., M.P., M.E., A.K., S.P., P.C., D.H., R.D., S.L.-M., and A.M.) have financial and personal relationships with other people or organizations that could inappropriately influence their work. Acknowledgement The authors would like to acknowledge the contribution of Ms. Mie Konno for collecting the data used in this study, and Dr. R. Wood for his helpful discussions. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.radonc.2010.08.019. References [1] Ferlay J, Bray F, Pisani P, et al. GLOBOCAN 2002: Cancer incidence, mortality and prevalence worldwide IARC CancerBase No. 5. version 2.0. Lyon: IARCPress; 2004. [2] Vale C, Tierney JF, Stewart LA, et al. Chemoradiotherapy for cervical cancer metaanalysis collaboration. Reducing uncertainties about the effects of chemoradiotherapy for cervical cancer: a systematic review and meta-analysis

C.M. Doll et al. / Radiotherapy and Oncology 97 (2010) 352–359

[3]

[4]

[5] [6]

[7] [8] [9]

[10]

[11] [12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

of individual patient data from 18 randomized trials. J Clin Oncol 2008;26: 5802–12. Green J, Kirwan J, Tierney J, Vale C, Symonds P, Fresco L, et al. Concomitant chemotherapy and radiation therapy for cancer of the uterine cervix. Cochrane Database Syst Rev 2005 [CD00225.pub2]. Sijbers AM, deLaat WL, Ariza RR, Biggerstaff M, Wei YF, Moggs JG, et al. Xeroderma pigmentosum group F caused by a defect in a structure-specific DNA repair endonuclease. Cell 1996;86:811–22. Enzlin JH, Scharer OD. The active site of the DNA repair endonuclease XPFERCC1 forms a highly conserved nuclease motif. EMBO J 2002;21:2045–53. Tsodikov OV, Enzlin JH, Scharer OD, Ellenberger T. Crystal structure and DNA binding functions of ERCC1, a subunit of the DNA structure-specific endonuclease XPF-ERCC1. Proc Natl Acad Sci USA 2005;102:11236–41. Gaillard PHL, Wood RD. Activity of individual ERCC1 and XPF subunits in DNA nucleotide excision repair. Nucleic Acids Res 2001;29:872–9. Gossage L, Madhusudan S. Current status of excision repair cross complementing-group 1 (ERCC1) in cancer. Cancer Treat Rev 2007;33:565–77. Adair GR, Rolig RL, Moore-Faver D, Zabelshansky M, Wilson JH, Nairn RS. Role of ERCC1 in removal of long non-homologous tails during targeted homologous recombination. EMBO J 2000;19:5552–61. Niedernhofer LJ, Essers J, Weeda G, Beverloo B, de Wit J, Muijtjens M, et al. The structure-specific endonuclease Ercc1-xpf is required for targeted gene replacement in embryonic stem cells. EMBO J 2001;20:6540–9. Bergstralh DT, Sekelsky J. Interstrand crosslink repair: can XPF-ERCC1 be let off the hook? Trends Genet 2008;24:70–6. Ahmad A, Robinson AR, Duensing A, van Drunen E, Beverloo HB, Weisberg DB, et al. ERCC1-XPF endonuclease facilitates DNA double-strand break repair. Mol Cell Biol 2008;28:5082–92. Shirota Y, Stoehlmacher J, Brabender J, Xiang YP, Uetake H, Danenberg KD, et al. ERCC1 and thymidylate synthase mRNA levels predict survival for colorectal cancer patients receiving combination oxaliplatin and fluorouracil chemotherapy. J Clin Oncol 2001;19:4298–304. Metzger R, Leichman CG, Danenberg KD, Danenberg PV, Lenz HJ, Hayashi K, et al. ERCC1 mRNA levels complement thymidylate synthase mRNA levels in predicting response and survival for gastric cancer patients receiving combination cisplatin and fluorouracil chemotherapy. J Clin Oncol 1998;16:309–16. Kwon HC, Roh MS, Oh SY, Kim SH, Kim MC, Kim JS, et al. Prognostic value of expression of ERCC1, thymidylate synthase, and glutathione S-transferase P1 for 5-fluorouracil/oxaliplatin chemotherapy in advanced gastric cancer. Ann Oncol 2007;18:504–9. Dabholkar M, Bostickbruton F, Weber C, Bohr VA, Egwuagu C, Reed E. Ercc1 and Ercc2 expression in malignant-tissues from ovarian-cancer patients. J Natl Cancer Inst 1992;84:1512–7. Warnecke-Eberz U, Metzger R, Miyazono F, Baldus SE, Neiss S, Brabender J, et al. High specificity of quantitative excision repair cross-complementing 1 messenger RNA expression for prediction of minor histopathological response to neoadjuvant radiochemotherapy in esophageal cancer. Clin Cancer Res 2004;10:3794–9. Handra-Luca A, Hernandez J, Mountzios G, Taranchon E, Lacau-St-Guily J, Soria J, et al. Excision repair cross complementation group 1 immunohistochemical expression predicts objective response and cancer-specific survival in patients treated by cisplatin-based induction chemotherapy for locally advanced head and neck squamous cell carcinoma. Clin Cancer Res 2007;13:3855–9. Jun HJ, Ahn MJ, Kim HS, Yi SY, Han J, Lee SK, et al. ERCC1 expression as a predictive marker of squamous cell carcinoma of the head and neck treated with cisplatin-based concurrent chemoradiation. Br J Cancer 2008;99:167–72. Lord RVN, Brabender J, Gandara D, Alberola V, Camps C, Domine M, et al. Low ERCC1 expression correlates with prolonged survival after cisplatin plus gemcitabine chemotherapy in non-small cell lung cancer. Clin Cancer Res 2002;8:2286–91. Ceppi P, Volante M, Novello S, Rapa I, Danenberg KD, Danenberg PV, et al. ERCC1 and RRM1 gene expressions but not EGFR are predictive of shorter survival in advanced non-small-cell lung cancer treated with cisplatin and gemcitabine. Ann Oncol 2006;17:1818–25. Olaussen KA, Dunant A, Fouret P, Brambilla E, Andre F, Haddad V, et al. DNA repair by ERCC1 in non-small-cell lung cancer and cisplatin-based adjuvant chemotherapy. N Engl J Med 2006;355:983–91. Kang CH, Jang BG, Kim DW, Chung DH, Kim YT, Jheon S, et al. The prognostic significance of ERCC1, BRCA1, XRCC1, and betaIII-tubulin expression in patients with non-small cell lung cancer treated by platinum- and taxanebased neoadjuvant chemotherapy and surgical resection. Lung Cancer 2010;68:478–83. Britten RA, Liu D, Tessier A, Hutchison MJ, Murray D. ERCC1 expression as a molecular marker of cisplatin resistance in human cervical tumor cells. Int J Cancer 2000;89:453–7. Khanna KK, Jackson SP. DNA double-strand breaks: signaling, repair, the cancer connection. Nat Genet 2001;27:247–54.

359

[26] Rubin JS, Whitmore GF. DNA repair-deficient chinese hamster ovary cells exhibiting differential sensitivity to gamma rays under aerobic, hypoxic conditions. Radiat Res 1985;101:528–34. [27] Murray D, Rosenberg E. The importance of the ERCC1/ERCC4[XPF] complex for hypoxic-cell radioresistance does not appear to derive from its participation in the nucleotide excision repair pathway. Mutat Res-DNA Repair 1996;364:217–26. [28] Hafer K, Iwamoto KK, Scuric Z, Schiestl RH. Adaptive response to gamma radiation in mammalian cells proficient, deficient in components of nucleotide excision repair. Radiat Res 2007;168:168–74. [29] Hampson L, El Hady ESA, Moore JV, Kitchener H, Hampson IN. The HPV16 E6 and E7 proteins and the radiation resistance of cervical carcinoma. FASEB J 2001;15:1445. [30] Carles J, Monzo M, Amat M, Jansa S, Artells R, Navarro A, et al. Singlenucleotide polymorphisms in base excision repair, nucleotide excision repair, and double strand break genes as markers for response to radiotherapy in patients with stage I to II head-and-neck cancer. Int J Radiat Oncol Biol Phys 2006;66:1022–30. [31] Niedernhofer LJ, Bhagwat N, Wood RD. ERCC1 and non-small-cell lung cancer. N Engl J Med 2007;356:2538–40. [32] Bhagwat NR, Roginskaya VY, Acquafondata MB, Dhir R, Wood RD, Niedernhofer LJ. Immunodetection of DNA repair endonuclease ERCC1-XPF in human tissue. Cancer Res 2009;69:6831–8. [33] Coulson DTR, Brockbank S, Quinn JG, Murphy S, Ravid R, Irvine GB, et al. Identification of valid reference genes for the normalization of RT qPCR gene expression data in human brain tissue. Bmc Mol Biol 2008;9:46. [34] Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 2002;3:23 [RESEARCH0034. Epub.]. [35] Andersen CL, Jensen JL, Orntoft TF. Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res 2004;64:5245–50. [36] Chung CH, Parker JS, Ely K, Carter J, Yi Y, Murphy BA, et al. Gene expression profiles identify epithelial-to-mesenchymal transition and activation of nuclear factor-kappa B signaling as characteristics of a high-risk head and neck squamous cell carcinoma. Cancer Res 2006;66:8210–8. [37] Livak KJ, Schmittgen TD. Analysis of relative gene expression data using realtime quantitative PCR, the 2(T)(-delta delta C) method. Methods 2001;25:402–8. [38] Camp RL, Chung GG, Rimm DL. Automated subcellular localization, quantification of protein expression in tissue microarrays. Nat Med 2002;8:1323–7. [39] Stehman FB, Bundy BN, Disaia PJ, Keys HM, Larson JE, Fowler WC. Carcinoma of the cervix treated with radiation therapy-i – a multi-variate analysis of prognostic variables in the gynecologic oncology group. Cancer 1991:2776–85. [40] Grogan M, Thomas GM, Melamed I, Wong FLW, Pearcey RG, Joseph PK, et al. The importance of hemoglobin levels during radiotherapy for carcinoma of the cervix. Cancer 1999;86:1528–36. [41] Murray D, Macann A, Hanson J, Rosenberg E. ERCC1/ERCC4 50 -endonuclease activity as a determinant of hypoxic cell radiosensitivity. Int J Radiat Biol 1996;69:319–27. [42] Simon GR, Sharma S, Cantor A, Smith P, Bepler G. ERCC1 expression is a predictor of survival in resected patients with non-small cell lung cancer. Chest 2005;127:978–83. [43] Olaussen KA, Fouret P, Kroemer G. ERCC1-specific immunostaining in nonsmall-cell lung cancer. N Engl J Med 2007;357:1559–61. [44] Zheng Z, Chen T, Li X, Haura E, Sharma A, Bepler G. DNA synthesis and repair genes RRM1 and ERCC1 in lung cancer. N Engl J Med 2007;356:800–8. [45] Goyal S, Parikh RR, Green C, Schiff D, Moran MS, Yang Q, et al. Clinicopathologic significance of excision repair cross-complementation 1 expression in patients treated with breast-conserving surgery and radiation therapy. Int J Radiat Oncol Biol Phys. 2009;May 21. Epub. [46] Griffin C, de Waard H, Deans B, Thacker J. The involvement of key DNA repair pathways in the formation of chromosome rearrangements in embryonic stem cells. DNA Repair 2005;4:1019–27. [47] Cheng L, Spitz MR, Hong WK, Wei QY. Reduced expression levels of nucleotide excision repair genes in lung cancer: a case-control analysis. Carcinogenesis 2000;21:1527–30. [48] Benchekroun M, DeGraw J, Gao J, Sun L, von Boguslawsky K, Leminen A, et al. Impact of fixative on recovery of mRNA from paraffin-embedded tissue. Diagn Mol Pathol 2004;13:116–25. [49] Cobo M, Isla D, Massuti B, Montes A, Miguel Sanchez J, Provencio M, et al. Customizing cisplatin based on quantitative excision repair crosscomplementing 1 mRNA expression: a phase III trial in non-small-cell lung cancer. JCO 2007:2747–54.