- Email: [email protected]
Tissue microarray constructs to predict a response to chemoradiation in rectal cancer
Digestive and Liver Disease 42 (2010) 679–684
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Alimentary Tract
Tissue microarray constructs to predict a response to chemoradiation in rectal cancer Sergio Huerta a,∗ , John Hrom b , Xiaohuan Gao a , Debabrata Saha c , Thomas Anthony a , Henry Reinhart a , Payal Kapur d a
Department of Surgery, University of Texas Southwestern Medical Center, United States Department of Medical Oncology, University of Texas Southwestern Medical Center, United States c Department of Radiation Oncology, University of Texas Southwestern Medical Center, United States d Department of Pathology, University of Texas Southwestern Medical Center, United States b
a r t i c l e
i n f o
Article history: Received 24 November 2009 Accepted 2 February 2010 Available online 15 March 2010 Keywords: BAX Bcl-2 MIB Number of lymph nodes Pathological complete response p53
a b s t r a c t Purpose: To identify, using tissue microarray (TMA), an immunohistochemical panel predictive of response to ionizing radiation (IR) in rectal cancer. Methods: TMA constructs were prepared from archived stage II/III rectal tumors and matching adjacent mucosa (n = 38) from patients treated with pre-operative chemoradiation. Immunohistochemistry (IHC) was performed for MIB, Cyclin E, p21, p27, p53, survivin, Bcl-2, and BAX. Immunoreactivity along with clinical variables was subjected to univariate and forward stepwise logistic regression analyses. Results: Pathological complete response (pCR) was 23.9%. The number of positive lymph nodes obtained in the resected specimen was associated with pCR. Immunoreactivity for MIB (Sn 15%, Sp 65%, OR 0.33), p53 (Sn 3%, Sp 84%, OR 0.16), Bcl-2 (Sn 11%, Sp 74%, OR 0.35), and BAX (Sn 92%, Sp 80%, OR 46) was associated with pathological response (all p’s < 0.001). Forward stepwise logistic regression analysis demonstrated that MIB was an independent predictor of a response to chemoradiation (p = 0.001). Conclusions: A combined panel of mediators of apoptosis alone or combined with clinical factors is a feasible approach that can be applied to rectal tumor biopsies to predict a response to chemoradiation. The most sensitive factor was BAX; while MIB independently predicted a response to chemoradiation. Published by Elsevier Ltd on behalf of Editrice Gastroenterologica Italiana S.r.l.
1. Introduction Pre-operative chemoradiotherapy for the management of stage II/III rectal cancer results in a wide spectrum in clinical response. Current data demonstrate a reduction in local recurrence [1]; however, survival advantage is uncertain [2–5]. Additionally, the ability of neoadjuvant therapy to reduce tumor size, a major objective of this pre-operative treatment, is extraordinarily unpredictable. Nine to 37% of patients demonstrate a pathological complete response (pCR) with the various chemoradiotherapeutic regimens currently available (i.e. 5-FU, ironotecan, oxaliplatin, bevacizumab, and cetuximab in combination with IR) [6]. On the other hand, up to 9% of these patients may not respond at all. Appropriate selection of patients in either side of the spectrum has not yet been possible either clinically or at the molecular level. The identification of patients at risk of being radioresistant would benefit clinicians and patients alike in determining individuals that would respond to this
∗ Corresponding author at: Dallas VA Medical Center, Department of Surgery (112), 4500 S. Lancaster Road, Dallas, TX 75216, United States. Tel.: +1 214 857 1800; fax: +1 214 648 6700. E-mail address: [email protected] (S. Huerta).
neoadjuvant modality. In the patients predicted to have radiosensitive tumors a selective and individualized form of chemoradiation could be instituted. In patients predicted to have radioresistant tumors, it is not justifiable to subject them to the adverse side effects of chemoradiation and alternative options could be undertaken such as surgical intervention without chemoradiation or inclusion of patients in clinical trials. Identification of a panel of specific molecular phenotype could be the cornerstone of chemoradiotherapeutic interventions. However, to date, no markers are available to predict tumor response to radiotherapy. Our current understanding of the mechanisms that lead to resistance to chemoradiation are limited. Various markers including p53 [7], p21 [8], p27 [9], NFB [10], survivin [11], Bcl-2 [12] and mediators of apoptosis [13], have been investigated either individually or in combination, with variable results as predictors of response to chemoradiation [reviewed in [6]]. The major limitation of these studies has been the use of individual markers by immunohistochemistry (IHC) in a heterogeneous patient population. In the present study, we have developed Tissue Microarray (TMA) constructs of tumor and corresponding normal mucosa from rectal cancer specimens of 36 patients who received pre-operative chemoradiation. IHC was performed with eight different antibodies. The aims of this study were to correlate the expression of these
1590-8658/$36.00 Published by Elsevier Ltd on behalf of Editrice Gastroenterologica Italiana S.r.l. doi:10.1016/j.dld.2010.02.003
680
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
markers either individually or in combination with clinical and oncologic outcome and to develop an immunohistochemical panel that could be used as a predictor of clinical response to chemoradiation. 2. Methods
micrometer-precise coordinate system for assembling tissue samples on a block. For each case, single 0.6 mm (diameter 0.6 mm, height 3–4 mm) core diameter samples were obtained from the circled areas of tumor and/or normal colonic mucosa from each “donor” block and placed on separate a “recipient” TMA block (45 mm × 20 mm, 0.7 mm center). All samples were spaced 0.5 mm apart as previously described [14].
2.1. Patients 2.4. Immunohistochemistry All studies were undertaken with the approval and institutional oversight of the Institutional Review Board (IRB) for the Protection of Human Subjects at the Dallas VA Medical (DVAMC) Center, the University of Texas Southwestern Medical Center, and Parkland Memorial Hospital (PMH). Clinical data were collected retrospectively from 59 patents from the DVAMC and 58 from PMH. Only patients with stage II and stage III rectal cancer who were subjected to neoadjuvant chemoradiation were included in this study. Per protocol, all our patients received 50.4 Gy ionizing radiation given in combination with oral capecitibine 825 mg/m2 two times daily, Monday through Friday, over 6 weeks. Pathological response was defined by comparing the pretreatment tumor size (as determined by EUS, CT, flexible sigmoidoscopy, and/or colonoscopy) with post-resection tumor size (as assessed by the pathological report and review of all slides from each specimen). 2.2. Tumor tissues In order to provide uniformity of pre-operative treatment and tissue handling, for this study, samples (n = 38) were collected from patients treated at a single institution (PMH) between April 2000 and November 2008. 2.3. Tissue microarray (TMA) All Hematoxylin and eosin (H&E) stained section from each specimen were reviewed by a staff pathologist (PK) to select representative areas of the tumor from which to acquire cores for microarray analysis. Two samples, for each patient, one from the tumor and one from normal colonic epithelium not adjacent to the cancer, were identified and circled on the H&E stained slides. Tissue microarrays were built using a semiautomatic arraying instrument (Beecher Instruments, Silver Spring, MD) that uses two separate core needles for punching the donor and recipient blocks and a
We performed immunohistochemical staining for p21 (monoclonal mouse, SX118, Dako; dilution 1:200), p53 (monoclonal mouse, DO-7, Dako; dilution 1:2200), p27 (monoclonal mouse, SX53G8, Dako; dilution 1:150), Bcl-2 (monoclonal mouse, 124, Dako; dilution 1:600), survivin (polyclonal rabbit, 8E 2, BioCare; dilution 1:300), Ki-67 (monoclonal mouse, MIB-1, Dako; dilution 1:300), Cyclin E (monoclonal mouse, 870P110, Dako; dilution 1:400) and BAX (monoclonal rabbit, A3533, Dako; dilution 1:200) using known positive control tissues. Appropriate positive and negative controls were used. All immunostaining was performed as we have previously described [14]. 2.5. TUNEL Terminal deoxynucleotidyltransferase-mediated UTP end labeling (TUNEL) staining for apoptotic cells was done on paraffinembedded TMA sections, according to the protocol supplied with Promega Apoptosis and as published before [15,16]. 2.6. Statistical analysis All data is expressed as means ±SE. PRISM statistical analysis software (GraphPad Software Inc., San Diego, CA) was used for contingency table analysis and Student’s T-test. SigmaPlot for windows version 11.0 (Systat Software Inc., San Jose, CA) was employed for multivariate analysis. 2.7. Univariate analysis Univariate analysis was performed in 117 patients by dividing the patient population into those who achieved a pCR (n = 28) and all others (n = 89). Categorical data was analyzed by Fisher’s Exact Test or Chi Square. Continuous data was subjected to Student’s Ttest. For analysis of the TMA constructs immunoreactivity in the
Fig. 1. There is a high variability of a response to ionizing radiation in rectal cancer. Each bar on the X-axis represents an individual patient. The Y-axis represents the clinical response to pre-operative ionizing radiation (the size of the tumor at diagnosis was compared to the pathological report size).
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
two groups: good responders (>50% pathological response; n = 19) vs. poor responders (<50% pathological response; n = 19) was also analyzed by contingency tables. 2.8. Multivariate analysis model Multivariate analysis was performed first in all clinical variables in 117 patients with pCR as the dependent variable and all of the clinical factors as independent variables by forward multistep logistic regression analysis. A second model was constructed between good responders and poor responders for analysis of immunoreactivity with pathological response as the dependent variable. Independent variables were all factors with a p-value of ≤0.2 (i.e. MIB from tumor, p21 from tumor, Bcl-2 from tumor, p53 from tumor, BAX from tumor, as well as clinical variables including number of positive nodes and tumor stage). This analysis was also performed by forward multistep logistic regression analysis. 3. Results A review of 117 patients who received pre-operative ionizing radiation at the DVAMC and PMH that had complete information regarding pre-operative assessment of tumor size revealed a wide range of pathological response. Over 23% of patients achieved a pathological complete response (n = 28). On the other side of the spectrum, 20.5% (n = 24) either had no response or the tumor continued to grow up to 200% of the original size. The rest of the patients had a pathological response in between (44.4%, n = 44; Fig. 1). The demographics for this patient population are depicted in Table 1. 3.1. Analysis of clinical variables Univariate analysis with pCR as the dependent variable and independent variables including: pre-operative CEA, race, gender, number of positive nodes, size of the tumor (T), and overall stage
681
Table 1 Patient demographics of all patients with stage II/III rectal cancers treated with neoadjuvant chemoradiation. Demographics n = 117 CpR = 23.9% (n = 28) Age = 59.8 ± 1.1-year-old Preop tumor size = 5.9 ± 0.3 cm Postop tumor size = 2.7 ± 0.2 cm White = 52% Hispanic = 22% Male = 75% Stage II = 60.7% (n = 71)
demonstrated that the only clinical parameter predictive of pCR was the number of positive nodes obtained from the resected specimen [0.07 ± 0.05 vs. 1.0 ± 0.24; p = 0.049 (T-test)]. While not statistically significant, stage demonstrated a trend towards statistic significance in patients with pCR [stage III (25.0%) vs. Stage II (43.2%), p = 0.07 (Chi square)]. None of these clinical variables demonstrated to be an independent predictor of pCR. We then proceeded to analyze tissue collected from 46 subjects who had complete information on the pre-operative size of the tumor as well as sufficient tissue from the tumor and from adjacent mucosa. Eight patients were eliminated from this analysis because they had achieved a pCR and there was no tumor available for the TMA. We performed TMA constructs as depicted in Fig. 2. 3.2. TMA analysis of normal mucosa TMA analysis of normal mucosa demonstrated no apoptosis as determined by TUNEL in either good or poor responders. Table 2 shows the results of the antibodies from the IHC of the TMA derived from normal mucosa. None of the antibodies investigated in this TMA demonstrated a significant association with tumor response by univariate analysis. Because the p-value of all of these molecules
Fig. 2. Tissue microarray constructs. The area of the tumor was examined by a pathologist (PK) and a pouch biopsy performed as described in Section 2. Samples were placed in a paraffin-embedded block, where each dot represents a different patient. Thin sections were then cut and placed onto slides for IHC analysis.
682
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
Table 2 Results univariate analysis from the tissue microarray in tissue derived from normal mucosa adjacent to tumor tissue. MIB (i.e. MIB-1 = anti-Ki-67 antibody), Cyc E (i.e. Cyclin E), Sn = sensitivity, Sp = specificity, PPV = positive predictive value, NPV = negative predictive value, OR = odds ratio. TMA normal mucosa MIB
Cyc E
p21
Good responders Poor responders
14.7 ± 8.1% 15.6 ± 3.8%
10.8 ± 1.4% 13.2 ± 3.1%
12.8 ± 2.4% 19.0 ± 4.9%
p Sn Sp PPV NPV OR
1.000 15% 84% 48% 50% 0.94
0.828 11% 79% 45% 49% 0.83
0.215 13% 81% 40% 48% 0.64
p27 8.3 ± 1.1% 6.9 ± 1.5% 1.000 8% 93% 53% 50% 1.15
p53
Survivin
Bcl-2
BAX
0.0 0.0
0.0 0.0
32.0 ± 14.3% 23.5 ± 5.3%
52.5 ± 6.0% 59.3 ± 7.9%
1.000 – – – – –
1.000 – – – – –
.027 32% 76% 57% 52% 1.49
0.476 53% 41% 47% 47% 0.78
Table 3 Results univariate analysis from the tissue microarray in tissue derived from tumor tissue. MIB (i.e. MIB-1 = anti-Ki-67 antibody), Cyc E (i.e. Cyclin E), Sn = sensitivity, Sp = specificity, PPV = positive predictive value, NPV = negative predictive value, OR = odds ratio. TMA tumor tissue MIB
Cyc E
p21
p27
p53
Survivin
Bcl-2
BAX
Good responders Poor responders
14.8 ± 6.0% 35.3 ± 8.1%
18.2 ± 3.4% 18.9 ± 4.3%
7.5 ± 5.3% 14.6 ± 3.4%
21.4 ± 5.6% 18.7 ± 4.3%
2.9 ± 2.0% 16.4 ± 3.8%
41.6 ± 10.5% 44.1 ± 10.1%
10.9 ± 6.3% 25.8 ± 5.9%
91.6 ± 4.2% 20.1 ± 4.6%
p Sn Sp PPV NPV OR
<0.001 15% 65% 30% 43% 0.33
1.000 18% 81% 49% 50% 0.93
0.180 8% 85% 35% 48% 0.67
0.860 21% 81% 52% 51% 1.13
<0.001 3% 84% 16% 46% 0.16
.890 42% 56% 49% 49% 0.92
<0.001 11% 74% 30% 45% 0.35
<0.001 92% 80% 82% 90% 46
was >0.2, they were not included in the multivariate analysis. Sensitivity (Sn), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and odds ratio (OR) was included in this table for completion and to allow comparison of these values with the TMA constructs from the tumor tissue. 3.3. TMA analysis of tumor tissue TUNEL analysis of the TMA constructs from tissue derived from the tumor demonstrated a rate of apoptosis 65.2 ± 2.0 in the good response group compared to 60.5 ± 3.0 in the poor response group (p = 0.470). The results of univariate analysis for each individual molecule with the respective Sn, Sp, PPV, NPV and the OR are depicted in Table 3. Univariate analysis demonstrated a positive relationship between a poor response to IR and immunoreactivity with antibodies specific for BAX (4.6-fold). An inverse relationship was found with antibodies specific for MIB (2.4-fold), p53 (5.7-fold) and Bcl-2 (2.4-fold). The highest sensitivity (92%) was for BAX with an OR of 46. Forward stepwise logistic regression analysis with response to IR as the dependent variable and immunoreactivity for MIB from tumor, p21 from tumor, Bcl-2 from tumor, p53 from tumor, BAX from tumor, as well as clinical variables including: number of positive nodes and tumor stage as the independent variables was performed demonstrated that MIB-Tumor (p = 0.001) was the only significant factor. 4. Discussion Factors that predict resistance to pre-operative chemoradiation remain at large. In the present study, we developed a TMA model with molecules classically associated with a good response to chemoradiation as assessed by a review of the literature in independent studies by IHC (for a review of these studies see Ref. [6]). The aim of this analysis was to determine whether we could use clinical variables along with TMA constructs derived directly from rectal tumors and normal adjacent mucosa to predict a response
to neoadjuvant treatment in patients receiving the same form of neoadjuvant modality for the management of stage II and III rectal cancer. In contrast to other studies, our analysis included several molecules (8 different antibodies and TUNEL) in the same TMA model along with clinical variables found to be associated with a pCR. Additionally, we investigated mucosa adjacent to the tumor and subjected this to statistical analysis within the same model. A previously published study has undertaken a similar approach by TMA in patients with rectal cancer treated with chemoradiation to evaluate recurrence and survival [17]. In their analysis, Debucquoy et al. demonstrated that Cox-2 emerged as a predictor of survival. Their study also found that EGFR, proliferation and apoptosis was affected by chemoradiation [17]. Our approach differed from this analysis in that our major goal was to predict tumors likely to respond to chemoradiation. This is important because our analysis showed that up to 21% of patients will not respond to this form of neoadjuvant treatment (Fig. 1). Of the molecules most commonly associated with a response to IR, p53 and the cyclin kinase dependent inhibitors (CDKI’s) such as p27 and p21 are commonly studied in rectal cancer. In our analysis, we combined p53 along with CDKIs in the same TMA from both normal mucosa and tissue derived from the tumor. Analysis from normal mucosa revealed no immunoreactivity for p53 and there was no difference with regards to the CDKIs in this TMA. The CDKI’s were not differentially expressed in good responders compared to poor responders in tissue derived from the tumor either. Our results demonstrated that p53 was differentially expressed between patients who achieved at least 50% of response to chemoradiation in the neoadjuvant setting in samples derived directly from the tumor. In ex vivo studies, p53 positivity by IHC signifies the mutated status of the protein because conformational changes of the p53 protein resulting from mutations lead to protein stability and a longer half-life, which allows for increased detection by the antibodies [18,19]. Previous studies have shown that nuclear expression of p53 in rectal cancers predicted treatment failure
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
and expression of nuclear p53 protein by (IHC) correlated with resistance to pre-operative chemoradiation [20]. Multiple studies have indicated p53 to be pivotal in radiationinduced apoptosis in colorectal cancer in vitro and in vivo [6], but these finding have been in disagreement with other analyses [21,22]. Furthermore, p53 mutations may render cells a more radiosensitive phenotype [23]. For instance, mutant p53 indicated resistance to apoptosis in rectal cancers compared to wild-type rectal tissues by IHC analysis [24]. Additionally, the mutational status of p53 has been shown to be an important factor to determine the functional status of this protein in terms of a response to the cytotoxic effects of chemoradiation. While most mutations are localized to exons 5–8 of the p53 gene, it is the mutation of codon 288 in exon 8 that seems to affect rectal cancers and lead to a worse prognosis [25]. Thus, these specific mutations (but not all p53 mutations) may lead to a more resistant phenotype in patients with rectal cancer. Our results are in agreement with the studies that show an association between p53 mutated form (i.e. positive immunoreactivity by IHC) and resistance to chemoradiation in tissue derived directly from the tumor. However, stepwise logistic regression analysis failed to identify p53 as an independent predictor of a response to chemoradiation. We also studied the mitochondrial mediators of apoptosis: BAX and Bcl-2 as well as the inhibitor of apoptosis (IAP): survivin. The relative ratio of pro-apoptotic BAX and anti-apoptotic Bcl-2 determine release of cytochrome c from the intermitochondrial membrane, potentially augmenting the apoptotic response [26]. Studies assessing the role of Bcl-2 and BAX in resistance to ionizing radiation remain unclear [6]. Our analysis demonstrated that tissue derived from the tumors that had a good response to chemoradiation was associated with increased levels of BAX and a concomitant decrease in the levels of Bcl-2. The BAX to Bcl-2 ratio was 10.7-fold greater in the tumors that achieved a good response to chemoradiation compared to those that did not. Theoretically, this should have been associated with an increased rate of apoptosis in the good responder group. Apoptosis is a mechanism by which IR exerts its therapeutic response. Thus, defects in the apoptotic machinery render resistance to radiation therapy in rectal cancer [6]. For instance, Rodel’s group has shown that spontaneous apoptosis in pre-treatment biopsies was a good predictor of pathological response [27]. Similarly, a recent study assessed the role of both intrinsic and radiation-induced apoptosis as markers for prognosis in rectal cancer. This analysis included 1198 tumor samples from the Dutch Total Mesorectal Excision trial. Apoptosis was assessed by TUNEL in TMA blocks. The rate of recurrence in patients who received irradiation was 5% compared to 10% in patients who did not. Non-irradiated patients with high apoptosis had a decrease in local recurrence by 1.7-fold [28]. Because we did not study tumor tissue prior to chemoradiation, our analysis did investigate spontaneous apoptosis. We did not observe a significant difference in radiation-induced apoptosis in tissue derived from the tumors. This might be a reflection of the small sample size of our cohort of patients. Survivin is one of eight members of the IAPs, which are under the regulation of NFB and directly inhibit caspases 9, 3, and 7 [26]. Survivin has been shown to act as a constitutive radioresistant factor in colorectal cancer cells [6]. Short interfering RNA of the survivin gene increased apoptosis in reduced survival in colorectal cancer cells [6]. Survivin has been demonstrated to cause, mitotic arrest, cell cycle redistribution, and increase DNA double bond breaks [6]. Additionally, the 5-year survival of patients with survivin positive stage II colon cancer tumors was 41% lower than patients with survivin negative tumors by IHC studies [6]. In spite of this strong evidence regarding survivin, our TMA analysis did not find a relationship between survivin and a response to IR.
683
Rapidly proliferating cells typically respond more substantially to cell damage induced by ionizing radiation. Assessment of cell proliferation is possible with the Ki-67 equivalent MIB antibody, which identifies cycling cells in formalin fixed tissue. In preirradiated tumor biopsies from rectal cancer patients, tumor size, proliferating cell nuclear antigen (PCNA)/mitotic activity and Ki-67 predicted pCR to chemoradiation [29]. Ki-67 has also been shown to be reduced in tumors compared to biopsies of rectal specimens following chemoradiation, which is consistent with the observation that rapidly proliferating cells are more sensitive to chemoradiation [17]. Our analysis investigated MIB only in post-irradiated tumors and found that poor responders to IR had retained a high proliferating index. Furthermore, MIB was an independent predictor of a response to chemoradiation. An obvious limitation of our study is the small sample size. However, the goal of a panel predictive of a response to IR should provide information for individual patients to be clinically useful such that a small number of patients is desirable in determining statistical significance in this setting. A second limitation is that we investigated irradiated archived tissue as tumor tissue was more easily available for TMA constructs. To overcome this limitation, we also included adjacent normal mucosa, which should have received less pre-operative IR. However, tissue derived from adjacent mucosa failed to identify any tumor markers predictive of a response to IR. Pre-irradiated biopsies are required to overcome these limitations. A third limitation of our study is that we arbitrarily divided tumors that responded to ionizing radiation into those that had ≥50% response the those that had ≤50% response. However, a 30% response might still be clinically significant depending on the tumor size. This limitation could be overcome by TMA constructs in pre-irradiated biopsies that are able to achieve a pCR compared to non-responders utilizing the same molecular markers. The present report supports the feasibility for such analysis. In summary, our analysis indicated that p53, Bcl-2, BAX, and MIB were associated with a good response to ionizing radiation. MIB emerged as an independent predictor of a response to IR with a NPV of 43% and an OR of 0.33. While multiple studies have been undertaken to identify the clinical utility of any of these factors independently by IHC, our analysis undertook eight classically involved variables to predict a good response to IR by TMA. This approach allowed for a more standardized treatment for all tissue samples. Our results provide evidence for the feasibility of TMA constructs with several molecules to evaluate pre-irradiated tumor biopsies to predict a response to IR. Clinically, this information might be applicable to develop TMAs in tumor biopsies in patients prior to irradiation to select subjects that might respond to chemoradiation compared to patients who are unlikely to benefit from this preoperative chemotherapeutic intervention. Further, understanding pathways of chemoresistance will enhance the development of novel radiosensitizing modalities for the management of patients who do not respond to conventional chemotherapeutic interventions. Conflict of interest statement None declared.
List of abbreviations TMA, tissue microarray; IHC, immunohistochemistry; IR, ionizing radiation; NPV, negative predictive value; PPV, positive predictive value; pCR, pathological complete response.
684
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
Acknowledgements This work was made possible by funds from the Department of Surgery at the University of Texas Southwestern (Shannon Funds) and a Veterans Affairs VISN17 New Investigator Award. References [1] Wong RK, Tandan V, De SS, et al. Pre-operative radiotherapy and curative surgery for the management of localized rectal carcinoma. Cochrane Database Syst Rev 2007. CD002102. [2] Marks G, Mohiuddin M, Borenstein BD. Preoperative radiation therapy and sphincter preservation by the combined abdominotranssacral technique for selected rectal cancers. Dis Colon Rectum 1985;28:565–71. [3] Minsky BD, Cohen AM, Enker WE, et al. Sphincter preservation with preoperative radiation therapy and coloanal anastomosis. Int J Radiat Oncol Biol Phys 1995;31:553–9. [4] Gerard A, Buyse M, Nordlinger B, et al. Preoperative radiotherapy as adjuvant treatment in rectal cancer. Final results of a randomized study of the European Organization for Research and Treatment of Cancer (EORTC). Ann Surg 1988;208:606–14. [5] Goldberg PA, Nicholls RJ, Porter NH, et al. Long-term results of a randomised trial of short-course low-dose adjuvant pre-operative radiotherapy for rectal cancer: reduction in local treatment failure. Eur J Cancer 1994;30:1602–6. [6] Huerta S, Gao X, Saha D. Mechanisms of resistance to ionizing radiation in rectal cancer. Expert Rev Mol Diagn 2009;9:469–80. [7] Lowe SW, Schmitt EM, Smith SW, et al. p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature 1993;362:847–9. [8] Waldman T, Lengauer C, Kinzler KW, et al. Uncoupling of S phase and mitosis induced by anticancer agents in cells lacking p21. Nature 1996;381:713–6. [9] Esposito G, Pucciarelli S, Alaggio R, et al. P27kip1 expression is associated with tumor response to preoperative chemoradiotherapy in rectal cancer. Ann Surg Oncol 2001;8:311–8. [10] Russo SM, Tepper JE, Baldwin Jr AS, et al. Enhancement of radiosensitivity by proteasome inhibition: implications for a role of NF-kappaB. Int J Radiat Oncol Biol Phys 2001;50:183–93. [11] Rodel F, Frey B, Leitmann W, et al. Survivin antisense oligonucleotides effectively radiosensitize colorectal cancer cells in both tissue culture and murine xenograft models. Int J Radiat Oncol Biol Phys 2008;71:247–55. [12] Bhatavdekar JM, Patel DD, Ghosh N, et al. Coexpression of Bcl-2, c-Myc, and p53 oncoproteins as prognostic discriminants in patients with colorectal carcinoma. Dis Colon Rectum 1997;40:785–90. [13] Meyn RE, Stephens LC, Milas L. Programmed cell death and radioresistance. Cancer Metastasis Rev 1996;15:119–31.
[14] Huerta S, Baay-Guzman G, Gonzalez-Bonilla CR, et al. In vitro and in vivo sensitization of SW620 metastatic colon cancer cells to CDDP-induced apoptosis by the nitric oxide donor DETANONOate: Involvement of AIF. Nitric Oxide 2009;20:182–94. [15] Bauer KD, Torre-Bueno J, Diel IJ, et al. Reliable and sensitive analysis of occult bone marrow metastases using automated cellular imaging. Clin Cancer Res 2000;6:3552–9. [16] Lu JR, Bassel-Duby R, Hawkins A, et al. Control of facial muscle development by MyoR and capsulin. Science 2002;298:2378–81. [17] Debucquoy A, Goethals L, Libbrecht L, et al. Molecular and clinico-pathological markers in rectal cancer: a tissue micro-array study. Int J Colorectal Dis 2009;24:129–38. [18] Baas IO, van den Berg FM, Mulder JW, et al. Potential false-positive results with antigen enhancement for immunohistochemistry of the p53 gene product in colorectal neoplasms. J Pathol 1996;178:264–7. [19] Finlay CA, Hinds PW, Tan TH, et al. Activating mutations for transformation by p53 produce a gene product that forms an hsc70-p53 complex with an altered half-life. Mol Cell Biol 1988;8:531–9. [20] Adell G, Sun XF, Stal O, et al. p53 status: an indicator for the effect of preoperative radiotherapy of rectal cancer. Radiother Oncol 1999;51:169–74. [21] Hendry JH, Cai WB, Roberts SA, et al. p53 deficiency sensitizes clonogenic cells to irradiation in the large but not the small intestine. Radiat Res 1997;148:254–9. [22] Slichenmyer WJ, Nelson WG, Slebos RJ, et al. Loss of a p53-associated G1 checkpoint does not decrease cell survival following DNA damage. Cancer Res 1993;53:4164–8. [23] Ribeiro JC, Barnetson AR, Fisher RJ, et al. Relationship between radiation response and p53 status in human bladder cancer cells. Int J Radiat Biol 1997;72:11–20. [24] Hamada M, Fujiwara T, Hizuta A, et al. The p53 gene is a potent determinant of chemosensitivity and radiosensitivity in gastric and colorectal cancers. J Cancer Res Clin Oncol 1996;122:360–5. [25] Vidaurreta M, Maestro ML, Sanz-Casla MT, et al. Colorectal carcinoma prognosis can be predicted by alterations in gene p53 exons 5 and 8. Int J Colorectal Dis 2008;23:581–6. [26] Huerta S, Goulet EJ, Huerta-Yepez S, et al. Screening and detection of apoptosis. J Surg Res 2007;139:143–56. [27] Rodel C, Grabenbauer GG, Papadopoulos T, et al. Apoptosis as a cellular predictor for histopathologic response to neoadjuvant radiochemotherapy in patients with rectal cancer. Int J Radiat Oncol Biol Phys 2002;52:294–303. [28] de Bruin EC, van de Velde CJ, van de Pas S, et al. Prognostic value of apoptosis in rectal cancer patients of the dutch total mesorectal excision trial: radiotherapy is redundant in intrinsically high-apoptotic tumors. Clin Cancer Res 2006;12:6432–6. [29] Willett CG, Warland G, Cheek R, et al. Proliferating cell nuclear antigen and mitotic activity in rectal cancer: predictor of response to preoperative irradiation. J Clin Oncol 1994;12:679–82.
Contents lists available at ScienceDirect
Digestive and Liver Disease journal homepage: www.elsevier.com/locate/dld
Alimentary Tract
Tissue microarray constructs to predict a response to chemoradiation in rectal cancer Sergio Huerta a,∗ , John Hrom b , Xiaohuan Gao a , Debabrata Saha c , Thomas Anthony a , Henry Reinhart a , Payal Kapur d a
Department of Surgery, University of Texas Southwestern Medical Center, United States Department of Medical Oncology, University of Texas Southwestern Medical Center, United States c Department of Radiation Oncology, University of Texas Southwestern Medical Center, United States d Department of Pathology, University of Texas Southwestern Medical Center, United States b
a r t i c l e
i n f o
Article history: Received 24 November 2009 Accepted 2 February 2010 Available online 15 March 2010 Keywords: BAX Bcl-2 MIB Number of lymph nodes Pathological complete response p53
a b s t r a c t Purpose: To identify, using tissue microarray (TMA), an immunohistochemical panel predictive of response to ionizing radiation (IR) in rectal cancer. Methods: TMA constructs were prepared from archived stage II/III rectal tumors and matching adjacent mucosa (n = 38) from patients treated with pre-operative chemoradiation. Immunohistochemistry (IHC) was performed for MIB, Cyclin E, p21, p27, p53, survivin, Bcl-2, and BAX. Immunoreactivity along with clinical variables was subjected to univariate and forward stepwise logistic regression analyses. Results: Pathological complete response (pCR) was 23.9%. The number of positive lymph nodes obtained in the resected specimen was associated with pCR. Immunoreactivity for MIB (Sn 15%, Sp 65%, OR 0.33), p53 (Sn 3%, Sp 84%, OR 0.16), Bcl-2 (Sn 11%, Sp 74%, OR 0.35), and BAX (Sn 92%, Sp 80%, OR 46) was associated with pathological response (all p’s < 0.001). Forward stepwise logistic regression analysis demonstrated that MIB was an independent predictor of a response to chemoradiation (p = 0.001). Conclusions: A combined panel of mediators of apoptosis alone or combined with clinical factors is a feasible approach that can be applied to rectal tumor biopsies to predict a response to chemoradiation. The most sensitive factor was BAX; while MIB independently predicted a response to chemoradiation. Published by Elsevier Ltd on behalf of Editrice Gastroenterologica Italiana S.r.l.
1. Introduction Pre-operative chemoradiotherapy for the management of stage II/III rectal cancer results in a wide spectrum in clinical response. Current data demonstrate a reduction in local recurrence [1]; however, survival advantage is uncertain [2–5]. Additionally, the ability of neoadjuvant therapy to reduce tumor size, a major objective of this pre-operative treatment, is extraordinarily unpredictable. Nine to 37% of patients demonstrate a pathological complete response (pCR) with the various chemoradiotherapeutic regimens currently available (i.e. 5-FU, ironotecan, oxaliplatin, bevacizumab, and cetuximab in combination with IR) [6]. On the other hand, up to 9% of these patients may not respond at all. Appropriate selection of patients in either side of the spectrum has not yet been possible either clinically or at the molecular level. The identification of patients at risk of being radioresistant would benefit clinicians and patients alike in determining individuals that would respond to this
∗ Corresponding author at: Dallas VA Medical Center, Department of Surgery (112), 4500 S. Lancaster Road, Dallas, TX 75216, United States. Tel.: +1 214 857 1800; fax: +1 214 648 6700. E-mail address: [email protected] (S. Huerta).
neoadjuvant modality. In the patients predicted to have radiosensitive tumors a selective and individualized form of chemoradiation could be instituted. In patients predicted to have radioresistant tumors, it is not justifiable to subject them to the adverse side effects of chemoradiation and alternative options could be undertaken such as surgical intervention without chemoradiation or inclusion of patients in clinical trials. Identification of a panel of specific molecular phenotype could be the cornerstone of chemoradiotherapeutic interventions. However, to date, no markers are available to predict tumor response to radiotherapy. Our current understanding of the mechanisms that lead to resistance to chemoradiation are limited. Various markers including p53 [7], p21 [8], p27 [9], NFB [10], survivin [11], Bcl-2 [12] and mediators of apoptosis [13], have been investigated either individually or in combination, with variable results as predictors of response to chemoradiation [reviewed in [6]]. The major limitation of these studies has been the use of individual markers by immunohistochemistry (IHC) in a heterogeneous patient population. In the present study, we have developed Tissue Microarray (TMA) constructs of tumor and corresponding normal mucosa from rectal cancer specimens of 36 patients who received pre-operative chemoradiation. IHC was performed with eight different antibodies. The aims of this study were to correlate the expression of these
1590-8658/$36.00 Published by Elsevier Ltd on behalf of Editrice Gastroenterologica Italiana S.r.l. doi:10.1016/j.dld.2010.02.003
680
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
markers either individually or in combination with clinical and oncologic outcome and to develop an immunohistochemical panel that could be used as a predictor of clinical response to chemoradiation. 2. Methods
micrometer-precise coordinate system for assembling tissue samples on a block. For each case, single 0.6 mm (diameter 0.6 mm, height 3–4 mm) core diameter samples were obtained from the circled areas of tumor and/or normal colonic mucosa from each “donor” block and placed on separate a “recipient” TMA block (45 mm × 20 mm, 0.7 mm center). All samples were spaced 0.5 mm apart as previously described [14].
2.1. Patients 2.4. Immunohistochemistry All studies were undertaken with the approval and institutional oversight of the Institutional Review Board (IRB) for the Protection of Human Subjects at the Dallas VA Medical (DVAMC) Center, the University of Texas Southwestern Medical Center, and Parkland Memorial Hospital (PMH). Clinical data were collected retrospectively from 59 patents from the DVAMC and 58 from PMH. Only patients with stage II and stage III rectal cancer who were subjected to neoadjuvant chemoradiation were included in this study. Per protocol, all our patients received 50.4 Gy ionizing radiation given in combination with oral capecitibine 825 mg/m2 two times daily, Monday through Friday, over 6 weeks. Pathological response was defined by comparing the pretreatment tumor size (as determined by EUS, CT, flexible sigmoidoscopy, and/or colonoscopy) with post-resection tumor size (as assessed by the pathological report and review of all slides from each specimen). 2.2. Tumor tissues In order to provide uniformity of pre-operative treatment and tissue handling, for this study, samples (n = 38) were collected from patients treated at a single institution (PMH) between April 2000 and November 2008. 2.3. Tissue microarray (TMA) All Hematoxylin and eosin (H&E) stained section from each specimen were reviewed by a staff pathologist (PK) to select representative areas of the tumor from which to acquire cores for microarray analysis. Two samples, for each patient, one from the tumor and one from normal colonic epithelium not adjacent to the cancer, were identified and circled on the H&E stained slides. Tissue microarrays were built using a semiautomatic arraying instrument (Beecher Instruments, Silver Spring, MD) that uses two separate core needles for punching the donor and recipient blocks and a
We performed immunohistochemical staining for p21 (monoclonal mouse, SX118, Dako; dilution 1:200), p53 (monoclonal mouse, DO-7, Dako; dilution 1:2200), p27 (monoclonal mouse, SX53G8, Dako; dilution 1:150), Bcl-2 (monoclonal mouse, 124, Dako; dilution 1:600), survivin (polyclonal rabbit, 8E 2, BioCare; dilution 1:300), Ki-67 (monoclonal mouse, MIB-1, Dako; dilution 1:300), Cyclin E (monoclonal mouse, 870P110, Dako; dilution 1:400) and BAX (monoclonal rabbit, A3533, Dako; dilution 1:200) using known positive control tissues. Appropriate positive and negative controls were used. All immunostaining was performed as we have previously described [14]. 2.5. TUNEL Terminal deoxynucleotidyltransferase-mediated UTP end labeling (TUNEL) staining for apoptotic cells was done on paraffinembedded TMA sections, according to the protocol supplied with Promega Apoptosis and as published before [15,16]. 2.6. Statistical analysis All data is expressed as means ±SE. PRISM statistical analysis software (GraphPad Software Inc., San Diego, CA) was used for contingency table analysis and Student’s T-test. SigmaPlot for windows version 11.0 (Systat Software Inc., San Jose, CA) was employed for multivariate analysis. 2.7. Univariate analysis Univariate analysis was performed in 117 patients by dividing the patient population into those who achieved a pCR (n = 28) and all others (n = 89). Categorical data was analyzed by Fisher’s Exact Test or Chi Square. Continuous data was subjected to Student’s Ttest. For analysis of the TMA constructs immunoreactivity in the
Fig. 1. There is a high variability of a response to ionizing radiation in rectal cancer. Each bar on the X-axis represents an individual patient. The Y-axis represents the clinical response to pre-operative ionizing radiation (the size of the tumor at diagnosis was compared to the pathological report size).
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
two groups: good responders (>50% pathological response; n = 19) vs. poor responders (<50% pathological response; n = 19) was also analyzed by contingency tables. 2.8. Multivariate analysis model Multivariate analysis was performed first in all clinical variables in 117 patients with pCR as the dependent variable and all of the clinical factors as independent variables by forward multistep logistic regression analysis. A second model was constructed between good responders and poor responders for analysis of immunoreactivity with pathological response as the dependent variable. Independent variables were all factors with a p-value of ≤0.2 (i.e. MIB from tumor, p21 from tumor, Bcl-2 from tumor, p53 from tumor, BAX from tumor, as well as clinical variables including number of positive nodes and tumor stage). This analysis was also performed by forward multistep logistic regression analysis. 3. Results A review of 117 patients who received pre-operative ionizing radiation at the DVAMC and PMH that had complete information regarding pre-operative assessment of tumor size revealed a wide range of pathological response. Over 23% of patients achieved a pathological complete response (n = 28). On the other side of the spectrum, 20.5% (n = 24) either had no response or the tumor continued to grow up to 200% of the original size. The rest of the patients had a pathological response in between (44.4%, n = 44; Fig. 1). The demographics for this patient population are depicted in Table 1. 3.1. Analysis of clinical variables Univariate analysis with pCR as the dependent variable and independent variables including: pre-operative CEA, race, gender, number of positive nodes, size of the tumor (T), and overall stage
681
Table 1 Patient demographics of all patients with stage II/III rectal cancers treated with neoadjuvant chemoradiation. Demographics n = 117 CpR = 23.9% (n = 28) Age = 59.8 ± 1.1-year-old Preop tumor size = 5.9 ± 0.3 cm Postop tumor size = 2.7 ± 0.2 cm White = 52% Hispanic = 22% Male = 75% Stage II = 60.7% (n = 71)
demonstrated that the only clinical parameter predictive of pCR was the number of positive nodes obtained from the resected specimen [0.07 ± 0.05 vs. 1.0 ± 0.24; p = 0.049 (T-test)]. While not statistically significant, stage demonstrated a trend towards statistic significance in patients with pCR [stage III (25.0%) vs. Stage II (43.2%), p = 0.07 (Chi square)]. None of these clinical variables demonstrated to be an independent predictor of pCR. We then proceeded to analyze tissue collected from 46 subjects who had complete information on the pre-operative size of the tumor as well as sufficient tissue from the tumor and from adjacent mucosa. Eight patients were eliminated from this analysis because they had achieved a pCR and there was no tumor available for the TMA. We performed TMA constructs as depicted in Fig. 2. 3.2. TMA analysis of normal mucosa TMA analysis of normal mucosa demonstrated no apoptosis as determined by TUNEL in either good or poor responders. Table 2 shows the results of the antibodies from the IHC of the TMA derived from normal mucosa. None of the antibodies investigated in this TMA demonstrated a significant association with tumor response by univariate analysis. Because the p-value of all of these molecules
Fig. 2. Tissue microarray constructs. The area of the tumor was examined by a pathologist (PK) and a pouch biopsy performed as described in Section 2. Samples were placed in a paraffin-embedded block, where each dot represents a different patient. Thin sections were then cut and placed onto slides for IHC analysis.
682
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
Table 2 Results univariate analysis from the tissue microarray in tissue derived from normal mucosa adjacent to tumor tissue. MIB (i.e. MIB-1 = anti-Ki-67 antibody), Cyc E (i.e. Cyclin E), Sn = sensitivity, Sp = specificity, PPV = positive predictive value, NPV = negative predictive value, OR = odds ratio. TMA normal mucosa MIB
Cyc E
p21
Good responders Poor responders
14.7 ± 8.1% 15.6 ± 3.8%
10.8 ± 1.4% 13.2 ± 3.1%
12.8 ± 2.4% 19.0 ± 4.9%
p Sn Sp PPV NPV OR
1.000 15% 84% 48% 50% 0.94
0.828 11% 79% 45% 49% 0.83
0.215 13% 81% 40% 48% 0.64
p27 8.3 ± 1.1% 6.9 ± 1.5% 1.000 8% 93% 53% 50% 1.15
p53
Survivin
Bcl-2
BAX
0.0 0.0
0.0 0.0
32.0 ± 14.3% 23.5 ± 5.3%
52.5 ± 6.0% 59.3 ± 7.9%
1.000 – – – – –
1.000 – – – – –
.027 32% 76% 57% 52% 1.49
0.476 53% 41% 47% 47% 0.78
Table 3 Results univariate analysis from the tissue microarray in tissue derived from tumor tissue. MIB (i.e. MIB-1 = anti-Ki-67 antibody), Cyc E (i.e. Cyclin E), Sn = sensitivity, Sp = specificity, PPV = positive predictive value, NPV = negative predictive value, OR = odds ratio. TMA tumor tissue MIB
Cyc E
p21
p27
p53
Survivin
Bcl-2
BAX
Good responders Poor responders
14.8 ± 6.0% 35.3 ± 8.1%
18.2 ± 3.4% 18.9 ± 4.3%
7.5 ± 5.3% 14.6 ± 3.4%
21.4 ± 5.6% 18.7 ± 4.3%
2.9 ± 2.0% 16.4 ± 3.8%
41.6 ± 10.5% 44.1 ± 10.1%
10.9 ± 6.3% 25.8 ± 5.9%
91.6 ± 4.2% 20.1 ± 4.6%
p Sn Sp PPV NPV OR
<0.001 15% 65% 30% 43% 0.33
1.000 18% 81% 49% 50% 0.93
0.180 8% 85% 35% 48% 0.67
0.860 21% 81% 52% 51% 1.13
<0.001 3% 84% 16% 46% 0.16
.890 42% 56% 49% 49% 0.92
<0.001 11% 74% 30% 45% 0.35
<0.001 92% 80% 82% 90% 46
was >0.2, they were not included in the multivariate analysis. Sensitivity (Sn), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and odds ratio (OR) was included in this table for completion and to allow comparison of these values with the TMA constructs from the tumor tissue. 3.3. TMA analysis of tumor tissue TUNEL analysis of the TMA constructs from tissue derived from the tumor demonstrated a rate of apoptosis 65.2 ± 2.0 in the good response group compared to 60.5 ± 3.0 in the poor response group (p = 0.470). The results of univariate analysis for each individual molecule with the respective Sn, Sp, PPV, NPV and the OR are depicted in Table 3. Univariate analysis demonstrated a positive relationship between a poor response to IR and immunoreactivity with antibodies specific for BAX (4.6-fold). An inverse relationship was found with antibodies specific for MIB (2.4-fold), p53 (5.7-fold) and Bcl-2 (2.4-fold). The highest sensitivity (92%) was for BAX with an OR of 46. Forward stepwise logistic regression analysis with response to IR as the dependent variable and immunoreactivity for MIB from tumor, p21 from tumor, Bcl-2 from tumor, p53 from tumor, BAX from tumor, as well as clinical variables including: number of positive nodes and tumor stage as the independent variables was performed demonstrated that MIB-Tumor (p = 0.001) was the only significant factor. 4. Discussion Factors that predict resistance to pre-operative chemoradiation remain at large. In the present study, we developed a TMA model with molecules classically associated with a good response to chemoradiation as assessed by a review of the literature in independent studies by IHC (for a review of these studies see Ref. [6]). The aim of this analysis was to determine whether we could use clinical variables along with TMA constructs derived directly from rectal tumors and normal adjacent mucosa to predict a response
to neoadjuvant treatment in patients receiving the same form of neoadjuvant modality for the management of stage II and III rectal cancer. In contrast to other studies, our analysis included several molecules (8 different antibodies and TUNEL) in the same TMA model along with clinical variables found to be associated with a pCR. Additionally, we investigated mucosa adjacent to the tumor and subjected this to statistical analysis within the same model. A previously published study has undertaken a similar approach by TMA in patients with rectal cancer treated with chemoradiation to evaluate recurrence and survival [17]. In their analysis, Debucquoy et al. demonstrated that Cox-2 emerged as a predictor of survival. Their study also found that EGFR, proliferation and apoptosis was affected by chemoradiation [17]. Our approach differed from this analysis in that our major goal was to predict tumors likely to respond to chemoradiation. This is important because our analysis showed that up to 21% of patients will not respond to this form of neoadjuvant treatment (Fig. 1). Of the molecules most commonly associated with a response to IR, p53 and the cyclin kinase dependent inhibitors (CDKI’s) such as p27 and p21 are commonly studied in rectal cancer. In our analysis, we combined p53 along with CDKIs in the same TMA from both normal mucosa and tissue derived from the tumor. Analysis from normal mucosa revealed no immunoreactivity for p53 and there was no difference with regards to the CDKIs in this TMA. The CDKI’s were not differentially expressed in good responders compared to poor responders in tissue derived from the tumor either. Our results demonstrated that p53 was differentially expressed between patients who achieved at least 50% of response to chemoradiation in the neoadjuvant setting in samples derived directly from the tumor. In ex vivo studies, p53 positivity by IHC signifies the mutated status of the protein because conformational changes of the p53 protein resulting from mutations lead to protein stability and a longer half-life, which allows for increased detection by the antibodies [18,19]. Previous studies have shown that nuclear expression of p53 in rectal cancers predicted treatment failure
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
and expression of nuclear p53 protein by (IHC) correlated with resistance to pre-operative chemoradiation [20]. Multiple studies have indicated p53 to be pivotal in radiationinduced apoptosis in colorectal cancer in vitro and in vivo [6], but these finding have been in disagreement with other analyses [21,22]. Furthermore, p53 mutations may render cells a more radiosensitive phenotype [23]. For instance, mutant p53 indicated resistance to apoptosis in rectal cancers compared to wild-type rectal tissues by IHC analysis [24]. Additionally, the mutational status of p53 has been shown to be an important factor to determine the functional status of this protein in terms of a response to the cytotoxic effects of chemoradiation. While most mutations are localized to exons 5–8 of the p53 gene, it is the mutation of codon 288 in exon 8 that seems to affect rectal cancers and lead to a worse prognosis [25]. Thus, these specific mutations (but not all p53 mutations) may lead to a more resistant phenotype in patients with rectal cancer. Our results are in agreement with the studies that show an association between p53 mutated form (i.e. positive immunoreactivity by IHC) and resistance to chemoradiation in tissue derived directly from the tumor. However, stepwise logistic regression analysis failed to identify p53 as an independent predictor of a response to chemoradiation. We also studied the mitochondrial mediators of apoptosis: BAX and Bcl-2 as well as the inhibitor of apoptosis (IAP): survivin. The relative ratio of pro-apoptotic BAX and anti-apoptotic Bcl-2 determine release of cytochrome c from the intermitochondrial membrane, potentially augmenting the apoptotic response [26]. Studies assessing the role of Bcl-2 and BAX in resistance to ionizing radiation remain unclear [6]. Our analysis demonstrated that tissue derived from the tumors that had a good response to chemoradiation was associated with increased levels of BAX and a concomitant decrease in the levels of Bcl-2. The BAX to Bcl-2 ratio was 10.7-fold greater in the tumors that achieved a good response to chemoradiation compared to those that did not. Theoretically, this should have been associated with an increased rate of apoptosis in the good responder group. Apoptosis is a mechanism by which IR exerts its therapeutic response. Thus, defects in the apoptotic machinery render resistance to radiation therapy in rectal cancer [6]. For instance, Rodel’s group has shown that spontaneous apoptosis in pre-treatment biopsies was a good predictor of pathological response [27]. Similarly, a recent study assessed the role of both intrinsic and radiation-induced apoptosis as markers for prognosis in rectal cancer. This analysis included 1198 tumor samples from the Dutch Total Mesorectal Excision trial. Apoptosis was assessed by TUNEL in TMA blocks. The rate of recurrence in patients who received irradiation was 5% compared to 10% in patients who did not. Non-irradiated patients with high apoptosis had a decrease in local recurrence by 1.7-fold [28]. Because we did not study tumor tissue prior to chemoradiation, our analysis did investigate spontaneous apoptosis. We did not observe a significant difference in radiation-induced apoptosis in tissue derived from the tumors. This might be a reflection of the small sample size of our cohort of patients. Survivin is one of eight members of the IAPs, which are under the regulation of NFB and directly inhibit caspases 9, 3, and 7 [26]. Survivin has been shown to act as a constitutive radioresistant factor in colorectal cancer cells [6]. Short interfering RNA of the survivin gene increased apoptosis in reduced survival in colorectal cancer cells [6]. Survivin has been demonstrated to cause, mitotic arrest, cell cycle redistribution, and increase DNA double bond breaks [6]. Additionally, the 5-year survival of patients with survivin positive stage II colon cancer tumors was 41% lower than patients with survivin negative tumors by IHC studies [6]. In spite of this strong evidence regarding survivin, our TMA analysis did not find a relationship between survivin and a response to IR.
683
Rapidly proliferating cells typically respond more substantially to cell damage induced by ionizing radiation. Assessment of cell proliferation is possible with the Ki-67 equivalent MIB antibody, which identifies cycling cells in formalin fixed tissue. In preirradiated tumor biopsies from rectal cancer patients, tumor size, proliferating cell nuclear antigen (PCNA)/mitotic activity and Ki-67 predicted pCR to chemoradiation [29]. Ki-67 has also been shown to be reduced in tumors compared to biopsies of rectal specimens following chemoradiation, which is consistent with the observation that rapidly proliferating cells are more sensitive to chemoradiation [17]. Our analysis investigated MIB only in post-irradiated tumors and found that poor responders to IR had retained a high proliferating index. Furthermore, MIB was an independent predictor of a response to chemoradiation. An obvious limitation of our study is the small sample size. However, the goal of a panel predictive of a response to IR should provide information for individual patients to be clinically useful such that a small number of patients is desirable in determining statistical significance in this setting. A second limitation is that we investigated irradiated archived tissue as tumor tissue was more easily available for TMA constructs. To overcome this limitation, we also included adjacent normal mucosa, which should have received less pre-operative IR. However, tissue derived from adjacent mucosa failed to identify any tumor markers predictive of a response to IR. Pre-irradiated biopsies are required to overcome these limitations. A third limitation of our study is that we arbitrarily divided tumors that responded to ionizing radiation into those that had ≥50% response the those that had ≤50% response. However, a 30% response might still be clinically significant depending on the tumor size. This limitation could be overcome by TMA constructs in pre-irradiated biopsies that are able to achieve a pCR compared to non-responders utilizing the same molecular markers. The present report supports the feasibility for such analysis. In summary, our analysis indicated that p53, Bcl-2, BAX, and MIB were associated with a good response to ionizing radiation. MIB emerged as an independent predictor of a response to IR with a NPV of 43% and an OR of 0.33. While multiple studies have been undertaken to identify the clinical utility of any of these factors independently by IHC, our analysis undertook eight classically involved variables to predict a good response to IR by TMA. This approach allowed for a more standardized treatment for all tissue samples. Our results provide evidence for the feasibility of TMA constructs with several molecules to evaluate pre-irradiated tumor biopsies to predict a response to IR. Clinically, this information might be applicable to develop TMAs in tumor biopsies in patients prior to irradiation to select subjects that might respond to chemoradiation compared to patients who are unlikely to benefit from this preoperative chemotherapeutic intervention. Further, understanding pathways of chemoresistance will enhance the development of novel radiosensitizing modalities for the management of patients who do not respond to conventional chemotherapeutic interventions. Conflict of interest statement None declared.
List of abbreviations TMA, tissue microarray; IHC, immunohistochemistry; IR, ionizing radiation; NPV, negative predictive value; PPV, positive predictive value; pCR, pathological complete response.
684
S. Huerta et al. / Digestive and Liver Disease 42 (2010) 679–684
Acknowledgements This work was made possible by funds from the Department of Surgery at the University of Texas Southwestern (Shannon Funds) and a Veterans Affairs VISN17 New Investigator Award. References [1] Wong RK, Tandan V, De SS, et al. Pre-operative radiotherapy and curative surgery for the management of localized rectal carcinoma. Cochrane Database Syst Rev 2007. CD002102. [2] Marks G, Mohiuddin M, Borenstein BD. Preoperative radiation therapy and sphincter preservation by the combined abdominotranssacral technique for selected rectal cancers. Dis Colon Rectum 1985;28:565–71. [3] Minsky BD, Cohen AM, Enker WE, et al. Sphincter preservation with preoperative radiation therapy and coloanal anastomosis. Int J Radiat Oncol Biol Phys 1995;31:553–9. [4] Gerard A, Buyse M, Nordlinger B, et al. Preoperative radiotherapy as adjuvant treatment in rectal cancer. Final results of a randomized study of the European Organization for Research and Treatment of Cancer (EORTC). Ann Surg 1988;208:606–14. [5] Goldberg PA, Nicholls RJ, Porter NH, et al. Long-term results of a randomised trial of short-course low-dose adjuvant pre-operative radiotherapy for rectal cancer: reduction in local treatment failure. Eur J Cancer 1994;30:1602–6. [6] Huerta S, Gao X, Saha D. Mechanisms of resistance to ionizing radiation in rectal cancer. Expert Rev Mol Diagn 2009;9:469–80. [7] Lowe SW, Schmitt EM, Smith SW, et al. p53 is required for radiation-induced apoptosis in mouse thymocytes. Nature 1993;362:847–9. [8] Waldman T, Lengauer C, Kinzler KW, et al. Uncoupling of S phase and mitosis induced by anticancer agents in cells lacking p21. Nature 1996;381:713–6. [9] Esposito G, Pucciarelli S, Alaggio R, et al. P27kip1 expression is associated with tumor response to preoperative chemoradiotherapy in rectal cancer. Ann Surg Oncol 2001;8:311–8. [10] Russo SM, Tepper JE, Baldwin Jr AS, et al. Enhancement of radiosensitivity by proteasome inhibition: implications for a role of NF-kappaB. Int J Radiat Oncol Biol Phys 2001;50:183–93. [11] Rodel F, Frey B, Leitmann W, et al. Survivin antisense oligonucleotides effectively radiosensitize colorectal cancer cells in both tissue culture and murine xenograft models. Int J Radiat Oncol Biol Phys 2008;71:247–55. [12] Bhatavdekar JM, Patel DD, Ghosh N, et al. Coexpression of Bcl-2, c-Myc, and p53 oncoproteins as prognostic discriminants in patients with colorectal carcinoma. Dis Colon Rectum 1997;40:785–90. [13] Meyn RE, Stephens LC, Milas L. Programmed cell death and radioresistance. Cancer Metastasis Rev 1996;15:119–31.
[14] Huerta S, Baay-Guzman G, Gonzalez-Bonilla CR, et al. In vitro and in vivo sensitization of SW620 metastatic colon cancer cells to CDDP-induced apoptosis by the nitric oxide donor DETANONOate: Involvement of AIF. Nitric Oxide 2009;20:182–94. [15] Bauer KD, Torre-Bueno J, Diel IJ, et al. Reliable and sensitive analysis of occult bone marrow metastases using automated cellular imaging. Clin Cancer Res 2000;6:3552–9. [16] Lu JR, Bassel-Duby R, Hawkins A, et al. Control of facial muscle development by MyoR and capsulin. Science 2002;298:2378–81. [17] Debucquoy A, Goethals L, Libbrecht L, et al. Molecular and clinico-pathological markers in rectal cancer: a tissue micro-array study. Int J Colorectal Dis 2009;24:129–38. [18] Baas IO, van den Berg FM, Mulder JW, et al. Potential false-positive results with antigen enhancement for immunohistochemistry of the p53 gene product in colorectal neoplasms. J Pathol 1996;178:264–7. [19] Finlay CA, Hinds PW, Tan TH, et al. Activating mutations for transformation by p53 produce a gene product that forms an hsc70-p53 complex with an altered half-life. Mol Cell Biol 1988;8:531–9. [20] Adell G, Sun XF, Stal O, et al. p53 status: an indicator for the effect of preoperative radiotherapy of rectal cancer. Radiother Oncol 1999;51:169–74. [21] Hendry JH, Cai WB, Roberts SA, et al. p53 deficiency sensitizes clonogenic cells to irradiation in the large but not the small intestine. Radiat Res 1997;148:254–9. [22] Slichenmyer WJ, Nelson WG, Slebos RJ, et al. Loss of a p53-associated G1 checkpoint does not decrease cell survival following DNA damage. Cancer Res 1993;53:4164–8. [23] Ribeiro JC, Barnetson AR, Fisher RJ, et al. Relationship between radiation response and p53 status in human bladder cancer cells. Int J Radiat Biol 1997;72:11–20. [24] Hamada M, Fujiwara T, Hizuta A, et al. The p53 gene is a potent determinant of chemosensitivity and radiosensitivity in gastric and colorectal cancers. J Cancer Res Clin Oncol 1996;122:360–5. [25] Vidaurreta M, Maestro ML, Sanz-Casla MT, et al. Colorectal carcinoma prognosis can be predicted by alterations in gene p53 exons 5 and 8. Int J Colorectal Dis 2008;23:581–6. [26] Huerta S, Goulet EJ, Huerta-Yepez S, et al. Screening and detection of apoptosis. J Surg Res 2007;139:143–56. [27] Rodel C, Grabenbauer GG, Papadopoulos T, et al. Apoptosis as a cellular predictor for histopathologic response to neoadjuvant radiochemotherapy in patients with rectal cancer. Int J Radiat Oncol Biol Phys 2002;52:294–303. [28] de Bruin EC, van de Velde CJ, van de Pas S, et al. Prognostic value of apoptosis in rectal cancer patients of the dutch total mesorectal excision trial: radiotherapy is redundant in intrinsically high-apoptotic tumors. Clin Cancer Res 2006;12:6432–6. [29] Willett CG, Warland G, Cheek R, et al. Proliferating cell nuclear antigen and mitotic activity in rectal cancer: predictor of response to preoperative irradiation. J Clin Oncol 1994;12:679–82.