AQUA analysis of thymidylate synthase reveals localization to be a key prognostic biomarker in 2 large cohorts of colorectal carcinoma.
* Context.--Increased thymidylate synthase expression is a marker
for decreased survival in colorectal cancer.
Objective.--Thymidylate synthase localizes to both the nucleus and cytoplasm, but how the relationship of these expression levels affects colon cancer outcome has yet to be determined.
Design.--Using AQUA, we assessed prognosis of thymidylate synthase expression as a function of subcellular localization in 2 retrospective cohorts of colorectal carcinoma. We used the first cohort (n = 599) as a training set, subsequently validating optimal expression cut points in the second cohort (n = 447).
Results.--A significant association between decreased 5-year disease-specific survival and increased nuclear expression (16% decreased survival [72% to 56%] for the top 60% of nuclear-expressing tumors [P < .001]) and cytoplasmic expression (12% decreased survival [70% to 58%] for the top 54% of cytoplasmic-expressing tumors [P = .02]) was observed for the training set. A higher nuclear to cytoplasmic ratio also correlated significantly with decreased survival (15% decreased survival [66% to 51%] for the top 25% of tumors [P < .001]). Applying these findings to the validation set, as a function of time to recurrence, only the ratio (P = .03 [expression ratio]; P = .18 [nuclear]; P = .71 [cytoplasmic]) showed a significant association with decreased time to recurrence. Additionally, the expression ratio significantly added to the prognostic value given by the primary tumor pathologic classification and nodal status.
Conclusions.--These data suggest the relationship of nuclear to cytoplasmic thymidylate synthase expression, given as a ratio of continuous AQUA scores, to be a strong predictor of colon cancer survival.
Carcinoma (Care and treatment)
Cancer (Care and treatment)
Colorectal cancer (Diagnosis)
Colorectal cancer (Care and treatment)
Gene expression (Physiological aspects)
Genetic markers (Identification and classification)
Gustavson, Mark D.
Molinaro, Annette M.
Camp, Robert L.
Rimm, David L.
|Publication:||Name: Archives of Pathology & Laboratory Medicine Publisher: College of American Pathologists Audience: Academic; Professional Format: Magazine/Journal Subject: Health Copyright: COPYRIGHT 2008 College of American Pathologists ISSN: 1543-2165|
|Issue:||Date: Nov, 2008 Source Volume: 132 Source Issue: 11|
|Geographic:||Geographic Scope: United States Geographic Code: 1USA United States|
Thymidylate synthase (TS) catalyzes the reductive methylation of
deoxyuridylate in the pathway for the production of deoxythymidine
triphosphate, which is critical for DNA synthesis. (1) Thymidylate
synthase expression, as a determinant of sensitivity to
fluoropyrimidines, has been demonstrated in vitro, (2,3) and TS
expression in vivo may play an important role in determining tumor
response to 5-fluorouracil (5-FU). (4,5) Thymidylate synthase has been
suggested to be both prognostic (6,7) and predictive of response to
therapy (see Aschele et al (8) for review). However, considerable
heterogeneity exists in the percentage of positivity within the
population as well as variability in the literature about the overall
prognostic value of TS expression. (9) This variability is most likely
due to differences in methodologies, including differential definitions
of TS positivity determined by subjective assessments of expression
levels by traditional immunohistochemical techniques. Studies measuring
mRNA have removed considerable subjectivity but have still failed to
become part of the routine practice for management of colon cancer.
Recently, it has been demonstrated that TS may have other cellular functions, including translational regulation (see Liu et al (12) for review). Thus the subcellular localization of expression may be an important determinant of the functional role of TS. Because of the potential importance and functional consequence of TS subcellular localization, we wished to examine the role of TS localization as a function of survival. We recently developed a method of automated quantitative analysis (AQUA) that allows rapid analysis of immunofluorescence on tissue samples. (13) This method yields quantitative results comparable to enzyme-linked immunosorbent assays, but it can also measure levels of protein within user-defined subcellular compartments. (14) The method reduces the amount of human variability in scoring immunohistochemical staining by eye and results in a continuous range of protein expression scores rather than ordinal scales (0, + 1, +2, and +3) and has been used for a range of studies, including efforts to assess outcome as a function of the subcellular localization of expression of targets of interest. (15) Here, we examined TS expression, within nuclear and cytoplasmic compartments, on 2 independent cohorts representing more than 1000 colorectal cancer specimens in tissue microarray (TMA) format.
TMA Design and Processing
Tissue microarrays containing 599 and 447 primary colorectal carcinomas for training and validation cohorts, respectively (formalin-fixed, paraffin-embedded tumor samples; Yale University New Haven Hospital, New Haven, Conn, 1970 through 1981 for the training set; and from multiple sites, 1989 through 1996 for the validation set), were constructed at the Yale University (New Haven, Conn) and the University of Virginia (Charlottesville) TMA facilities, respectively. The validation set was the National Cancer Institute (NCI) colon cancer TMA, designed by statisticians at the NCI (Bethesda, Md), and intended for public distribution in an effort led by Dr Lisa McShane and others. Represented on the TMA are colon cancer specimens obtained from incident cases that occurred in members of the Kaiser Permanente Northwest Health Plan, 1989-1996.
Each tumor sample block was sectioned and first stained by hematoxylin-eosin, so that areas of invasive tumor could be identified and circled. The circled region was then transcribed onto the original block from which 0.6-mm cores were taken. Each core was arrayed into recipient blocks in a 1-mm-spaced grid, and 5-[micro]m-thick sections were cut and processed as previously reviewed and described. (16,17)
In brief, precut, paraffin-coated tissue microarray slides were deparaffinized and antigen retrieved by pressure cooking. Slides were preincubated with 0.3% bovine serum albumin in 0.1M Tris-buffered saline (TBS; pH 8.0) for 60 minutes at room temperature. Slides were then incubated with primary antibodies against TS (mouse monoclonal clone TS106, 1:100 dilution; LabVision NeoMarkers, Fremont, Calif) and pancytokeratin (rabbit polyclonal, 1:100 dilution; DAKO, Carpinteria, Calif) and diluted in bovine serum albumin/TBS overnight at 4[degrees]C. Slides were washed 3 times for 10 minutes with 1X TBS containing 0.05% Tween 20. Corresponding secondary antibodies were applied for 1 hour at room temperature in bovine serum albumin/TBS. These included either antibodies directly conjugated to a fluorophore for anti-cytokeratin (Alexa 488-conjugated goat anti-rabbit, 1:100; Molecular Probes, Eugene, Ore) and/or conjugated to a horseradish peroxidase for anti-thymidylate synthase (DAKO). Slides were again washed 3 times for 10 minutes with TBS containing 0.05% Tween 20. Slides were incubated with a fluorescent chromagen (Cy-5-tyramide, NEN Life Science Products, Boston, Mass), which, like diaminobenzidine, is activated by horseradish peroxidase and results in the deposition of numerous covalently associated Cy-5 dyes immediately adjacent to the horseradish peroxidase-conjugated secondary antibody. Cy-5 (red) was used because its emission peak is well outside the green-orange spectrum of tissue autofluorescence. Slides for automated analysis were coverslipped with an antifade 4'6-diamidino-2-phenylindole (DAPI)-containing mounting medium (ProLong Gold, Molecular Probes).
Automated image capture was performed by the AQUA system, which has previously been described in detail (18) and reviewed. (13,19,20) Using an Olympus BX51 microscope, images of the cytokeratin staining, visualized with Cy3, DAPI, and target staining with Cy5, were taken and saved for every histospot on the array. In-focus and out-of-focus images were taken for each channel for future use with the AQUA script and validation program.
AQUA analysis was performed as previously described. (13) In brief, a tumor-specific mask was generated by thresholding the image of a marker (cytokeratin) that differentiates tumor from surrounding stroma and/or leukocytes. This creates a binary mask (each pixel is either on or off). Thresholding levels were verified by spot checking a few images and then automated for the remaining images. All subsequent image manipulations involve only image information from the masked area. Next, 2 images (one in-focus, one slightly deeper) are taken of the compartment-specific tags and the target marker. A percentage of the out-of-focus image is subtracted from the in-focus image, based on a pixel-by-pixel analysis of the 2 images (using the rapid exponential subtraction algorithm). The algorithm also enhances the interface between areas of higher-intensity staining and adjacent areas of lower-intensity staining, allowing more accurate assignment of pixels of adjacent compartments. Finally, the pixel-based local assignment for compartmentalization of expression algorithm assigns each pixel in the image to a specific subcellular compartment. Pixels that cannot be accurately assigned to a compartment, within a user-defined degree of confidence, are discarded. For example, pixels in which the nuclear and cytoplasmic pixel intensities are too similar to be accurately assigned are negated (usually comprising <8% of the total pixels). Once each pixel is assigned to a subcellular compartment (or excluded as described above), the signal in each location is added up. These data are saved and can subsequently be expressed either as a percentage of the total signal or as the average signal intensity per compartment area. The score is expressed on a scale of 1 to 1000 as the total intensity detectable in a pixel range from 1 to 255, creating 3 significant figures. In this study, TS nuclear, cytoplasmic, and the ratio of nuclear to cytoplasmic signal were analyzed. Scores were adjusted according to amount of area covered by the subcellular compartments within the masked area.
Histospots containing <5% tumor, by mask area (automated), were excluded from further analysis. AQUA scores were normalized on a 0 to 100 scale for each cohort by dividing by the maximum AQUA score. For survival analysis, optimal cut points were selected using X-tile, as described previously. (21) Monte Carlo simulations were used to adjust for multiple looks in optimal cutpoint selection. (22) Hazard ratios were assessed using the univariate and multivariate Cox proportional hazards model (log-rank test at [alpha] = .05) using optimal cut points as determined by X-tile. Statistical analyses, including generation of Kaplan-Meier curves based on X-tile cut points, Cox regression, and linear regression models, were performed using SPSS v14.01 (SPSS, Inc, Chicago, Ill) and R (Free Software Foundation, Boston, Mass).
To quantitatively assess TS expression in colon cancer using AQUA, 2 large, independent, retrospective cohorts of colorectal carcinoma were obtained, each annotated with demographic, clinical, and follow-up information (Table 1). For the purpose of this study, the first cohort (n = 599; median disease-specific survival: 23 months) was used as a training set. The second cohort (n = 447; median recurrence-free survival: 20 months) was used as a validation set to corroborate the findings with the training set. Demographic and clinical makeup of each cohort is provided in Table 1.
AQUA takes advantage of the multiplexing power of fluorescence staining, which allows for staining of multiple markers on a single slide. In these experiments, each tumor sample was stained for TS (Cy5), cytokeratin to differentiate epithelium from stromal components as well as to identify cytoplasm (Cy2), and DAPI to distinguish nuclei. In Figure 1, A through F, staining patterns for each marker are given for 2 representative tumor samples. For each tumor sample, an AQUA score, which is directly proportional to molecules per unit area, was generated for TS expression in the nucleus and the cytoplasm. Figure 1, C, shows a tumor with high nuclear TS expression relative to cytoplasm (expression ratio, 1.54), whereas Figure 1, F, shows a tumor with lower nuclear expression relative to cytoplasm (expression ratio, 0.77).
[FIGURE 1 OMITTED]
An important consideration in quantitative assays such as these is experimental reproducibility. We, and others, have demonstrated that 2 tissue cores are representative of whole-tumor expression in >95% of cases. (23) To assess reproducibility, we stained separate, redundant cores for 152 of the 599 tumor samples in the training set and then performed regression analysis on the calculated AQUA scores. The resulting correlation coefficients provide an assessment of not only antibody/experimental reproducibility but also the expression heterogeneity. R values less than 0.4 are considered experimental failures, but R values between 0.4 and 0.7 would be considered indicative of heterogeneous marker expression, with those greater than 0.7 being considered more homogeneous. Figure 2 shows regression analysis between nuclear (R = 0.73; Spearman p = 0.74 [P < .001];Figure 2, A), cytoplasmic (R = 0.71; Spearman p = 0.73 [P < .001]; Figure 2, B), and the expression ratio (R = 0.79; Spearman [rho] = 0.77 [P < .001]; Figure 2, C). These results indicate high experimental re producibility, but they also indicate that TS expression within colon tumors is fairly homogenous. Figure 2, D, shows regression analysis between nuclear TS expression and the expression ratio (nuclear over cytoplasmic). The lack of correlation indicates that TS expression ratio does not correlate with nuclear expression level such that patients with low-level nuclear expression can still have a high expression ratio.
To observe the relationship between TS expression and patient outcome in a manner similar to that used for immunohistochemical data, but also in a rigorous manner for continuous data, we needed to define optimal cut points. We applied a recently developed statistical method called X-tile (21) to determine the optimal divisions of a continuous population. The optimal AQUA score cut point for nuclear TS expression on the training set was determined to be 27.4, which represents the top 60% of the population (Figure 3, A). Patients in this group had a 16% decrease (72% to 56%) in overall 5-year disease-specific survival. A significant outcome (P < .001) from the Monte Carlo simulation was observed for the optimal cut point, using 1000 randomly generated populations. However, that cut point was not significant when applied to the NCI validation set as a function of time to recurrence, even though the population segregation was comparable (Figure 3, B; P = .18). This suggests that nuclear TS expression is not a strong predictor of colon cancer outcome in all populations.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
Survival analysis of cytoplasmic TS expression in the training set (Figure 3, C) also revealed a significant association between increased expression and decreased 5-year disease-specific survival (12% decreased survival [70% to 58%] for the top 54% of cytoplasmic expressing tumors [Monte Carlo P = .02]). However, when this cut point was applied to the validation set with comparable population segregation, a significant association with survival was not observed (Figure 3, D; P = .71), thus suggesting that cytoplasmic TS expression is also not a strong predictor of outcome for colon cancer in all populations.
On the basis of data showing variable functional roles for TS in different subcellular compartments and taking advantage of the continuous nature of AQUA expression scores, we were able to generate a nuclear to cytoplasmic expression ratio for each tumor sample. Ratios were log transformed to normalize ratios less than 1 (shown here as actual ratios for ease of presentation), then analyzed as previously described. Tumors with high expression ratios, greater than 1.01 (top 25% of the population), showed a significant (Monte Carlo P < .001) 15% decrease (66% to 51%) in 5-year disease-specific survival on the training set (Figure 3, E). This cut point was validated (P = .03) in the second cohort (Figure 3, F), suggesting that a nuclear to cytoplasmic ratio is a strong predictor of outcome in at least 2 large independent populations of colon cancer patients.
To ascertain whether the nuclear to cytoplasmic ratio adds prognostic value in colon cancer with respect to other known clinical prognostic features, we investigated Cox proportional hazards, multivariate models on both the training and validation sets (Tables 2 and 3), first looking only at known clinical features common to both cohorts (T pathologic classification, nodal status, histologic grade, median age at diagnosis, and sex). On the training set, the best clinical model (Table 2) included T pathologic classification (hazard ratio [HR], 2.27; 95% confidence interval [95% CI], 1.56-3.29);P < .001), nodal status (HR, 3.55;95% CI, 2.38-5.39; P < .001), and sex (HR, 0.71; 95% CI, 0.520.96; P = .03). Histologic grade and age at diagnosis did not make significant contributions to the model (data not shown). Application of this model to the validation set demonstrated that only T pathologic classification (HR, 2.04; 95% CI, 1.01-4.13; P = .05) and nodal status (HR, 3.89; 95% CI, 2.24-6.77; P < .001), not sex (HR, 0.89; 95% CI, 0.60-1.34; P = .59) had significant prognostic value (Table 2). Using the covariates, T pathologic classification and nodal status, as our best overall clinical model, we examined the contribution of the TS expression ratio (Table 3). In this analysis, we used the optimal cut point to assign patients to 2 groups: those with a high (>1.01) ratio and those with a low (<1.01) ratio. For the training set, the expression ratio (HR, 1.79;95% CI, 1.30-2.67;P = .001) makes a significant contribution to the preestablished clinical model (Table 3). For the validation set, the addition of the TS expression ratio contributed prognostic significance (HR, 1.47; 95% CI, 0.94-2.28; P = .09) at the 10% level (Table 3).
To assess whether TS expression, as determined by AQUA analysis, can predict response to 5-FU treatment, we used a Cox proportional hazards model (data not shown) to assess the contribution of the TS nuclear to cytoplasmic ratio for prediction of outcome in a small subset of patients (n = 73) in the training set known to be treated with 5-FU. The model demonstrated that the ratio significantly predicted the outcome in the treatment group (HR, 2.89;95% CI, 1.12-7.47;P = .03). Furthermore, nuclear and cytoplasmic AQUA scores did not significantly predict outcome in the treatment group (data not shown). These results could not be validated in the independent cohort because of limited clinical information.
As has been seen previously, we were able to find prognostic value for TS expression in colon cancer. However, even using an objective and strictly quantitative approach, we found that neither the cytoplasmic nor the nuclear levels of TS could be validated as a prognostic marker in an independent cohort. However, we found that TS expression was a strong predictor of colon cancer outcome as a ratio of nuclear to cytoplasmic expression. We observed a 15% reduction in overall disease-specific survival in the training set, and then applied that expression ratio cut point to a second independent cohort, validating the result. Furthermore, given that we examined time to recurrence in the second cohort, these finding support that, not only does an expression ratio predict overall survival but also disease-free survival. Patients with a high expression ratio had a 17% reduction in recurrence-free survival. We also demonstrated, in a multivariate analysis, that the expression ratio adds prognostic significance to already existing clinical features used to predict overall and disease-free survival (T pathologic stage and nodal status). Thus, a TS expression ratio represents a novel prognostic biomarker that could potentially be used to influence decisions about the course of treatment for patients with colorectal cancer.
The novelty of these findings also stems from the expression ratio not being dependent on overall expression levels of TS (Figure 2, D). In the training set, 55% of patients in the high-expression-ratio group, showing decreased survival, had been characterized as having a better prognosis when based on total nuclear and/or cytoplasmic levels. Thus, the expression ratio provides a level of outcome prediction otherwise not afforded by measuring total cellular or subcellular levels of TS. This may be due to a number of factors, including the use of a ratio to normalize individual variability or artifacts in preparation or fixation. Furthermore, these findings support a hypothesis that it is the localization of TS within tumors that contributes to poorer disease outcome, not necessarily, total levels.
The primary role ascribed to TS is production of deoxythymidine triphosphate for DNA synthesis, a process largely considered to occur in the cytoplasm. (4,24,25) However, recent findings have shown TS to function in cellular proliferation and in RNA binding, where the protein acts as a translational repressor of several mRNAs, including p53 and c-myc (see Liu et al (12) for review). Although it remains unclear, nuclear localization of TS may be related to its RNA-binding function. This is supported by data showing that unbound/free TS, the form responsible for RNA binding, is predominantly localized in the nucleus. (26) Taken together with the data presented here, we could hypothesize that increased free TS (nuclear) relative to ternary or bound TS (cytoplasmic) is indicative of poorer outcome because of increased translational repression of key tumor suppressor genes, such as p53.
As mentioned previously, increased expression of TS has been associated with decreased response to 5-FU treatment. It has also been demonstrated that increased nuclear expression is associated with decreased response to therapy. (27) Preliminary evidence from our laboratory, using the training set, suggests that the TS-expression ratio, not nuclear or cytoplasmic expression alone, significantly predicts response to 5-FU treatment, as ascertained on a small subset (n = 73) of patients (data not shown; insufficient informational power to validate results on second cohort). If these data can be validated on a larger population of treated patients, they would demonstrate that patients with less available cytoplasmic TS relative to nuclear TS would have a decreased likelihood of treatment response.
Overall, these studies demonstrate that the nuclear to cytoplasmic expression ratio is a more powerful predictor of overall survival and disease-free survival in colorectal cancer patients than nuclear or cytoplasmic expression alone. As supported by multivariate analysis, this biomarker could be used with other common clinicopathologic criteria to better assess prognosis of patients in the clinic for determination of treatment course. With further study, this biomarker may also prove to be a potent, independent predictor for response to 5-FU treatment.
Accepted for publication May 6, 2008.
(1.) Santi DV. The mechanism and structure of thymidylate synthetase. Nucleic Acids Symp Ser (Oxf). 1986;17:125-126.
(2.) Berger SH, Berger FG. Thymidylate synthase as a determinant of 5-fluoro-2'-deoxyuridine response in human colonic tumor cell lines. Mol Pharmacol. 1988;34:474-479.
(3.) Johnston PG, Drake JC, Trepel J, Allegra CJ. Immunological quantitation of thymidylate synthase using the monoclonal antibody TS 106 in 5-fluorouracil-sensitive and -resistant human cancer cell lines. Cancer Res. 1992;52:4306 4312.
(4.) Johnston PG, Lenz HJ, Leichman CG, et al. Thymidylate synthase gene and protein expression correlate and are associated with response to 5-fluorouracil in human colorectal and gastric tumors. Cancer Res. 1995;55:1407-1412.
(5.) Leichman CG, Lenz HJ, Leichman L, et al. Quantitation of intratumoral thymidylate synthase expression predicts for disseminated colorectal cancer response and resistance to protracted-infusion fluorouracil and weekly leucovorin. J Clin Oncol. 1997;1 5:3223-3229.
(6.) Edler D, Glimelius B, Hallstrom M, et al. Thymidylate synthase expression in colorectal cancer: a prognostic and predictive marker of benefit from adjuvant fluorouracil-based chemotherapy. J Clin Oncol. 2002;20:1721-1728.
(7.) Edler D, Kressner U, Ragnhammar P, et al. Immunohistochemically detected thymidylate synthase in colorectal cancer: an independent prognostic factor of survival. Clin Cancer Res. 2000;6:488-492.
(8.) Aschele C, Lonardi S, Monfardini S. Thymidylate synthase expression as a predictor of clinical response to fluoropyrimidine-based chemotherapy in advanced colorectal cancer. Cancer Treat Rev. 2002;28:27-47.
(9.) Popat S, Matakidou A, Houlston RS. Thymidylate synthase expression and prognosis in colorectal cancer: a systematic review and meta-analysis. J Clin Oncol. 2004;22:529-536.
(10.) Leichman CG. Thymidylate synthase as a predictor of response. Oncology (Williston Park). 1998;12(suppl 6):43-47.
(11.) Leichman L, Lenz HJ, Leichman CG, et al. Quantitation of intratumoral thymidylate synthase expression predicts for resistance to protracted infusion of 5-fluorouracil and weekly leucovorin in disseminated colorectal cancers: preliminary report from an ongoing trial. Eur J Cancer. 1995;31A(7-8):1306-1310.
(12.) Liu J, Schmitz JC, Lin X, et al. Thymidylate synthase as a translational regulator of cellular gene expression. Biochim Biophys Acta. 2002;1587(2-3): 174-182.
(13.) Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat. Med. 2002;8: 1323-1327.
(14.) McCabe A, Dolled-Filhart M, Camp RL, Rimm DL. Automated quantitative analysis (AQUA) of in situ protein expression, antibody concentration, and prognosis. J Natl Cancer Inst. 2005;97:1808-1815.
(15.) Berger AJ, Kluger HM, Li N, et al. Subcellular localization of activating transcription factor 2 in melanoma specimens predicts patient survival 25. Cancer Res. 2003;63:8103-8107.
(16.) Rimm DL, Camp RL, Charette LA, Costa J, Olsen DA, Reiss M. Tissue microarray: a new technology for amplification of tissue resources. Cancer J. 2001;7:24-31.
(17.) Rimm DL, Camp RL, Charette LA, Olsen DA, Provost E. Amplification of tissue by construction of tissue microarrays. Exp Mol Pathol. 2001;70:255-264.
(18.) Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat Med. 2002;8: 1323-1327.
(19.) Giltnane JM, Rimm DL. Technology insight: identification of biomarkers with tissue microarray technology. Nat Clin Pract Oncol. 2004;1:104-111.
(20.) Cregger M, Berger AJ, Rimm DL. Immunohistochemistry and quantitative analysis of protein expression. Arch Pathol Lab Med. 2006;130:1026-1030.
(21.) Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization 9. Clin Cancer Res. 2004;10:7252-7259.
(22.) Raeside DE. Monte Carlo principles and applications. Phys Med Biol. 1976;21:181-197.
(23.) Camp RL, Charette LA, Rimm DL. Validation of tissue microarray technology in breast carcinoma 50. Lab Invest. 2000;80:1943-1949.
(24.) Kucera R, Paulus H. Localization of the deoxyribonucleotide biosynthetic enzymes ribonucleotide reductase and thymidylate synthase in mouse L cells. Exp Cell Res. 1986;167:417-428.
(25.) Johnston PG, Liang CM, Henry S, Chabner BA, Allegra CJ. Production and characterization of monoclonal antibodies that localize human thymidylate synthase in the cytoplasm of human cells and tissue. Cancer Res. 1991;51:6668 6676.
(26.) Bissoon-Haqqani S, Moyana T, Jonker D, Maroun JA, Birnboim HC. Nuclear expression of thymidylate synthase in colorectal cancer cell lines and clinical samples. J Histochem Cytochem. 2006;54:19-29.
(27.) Wong NA, Brett L, Stewart M, et al. Nuclear thymidylate synthase expression, p53 expression and 5FU response in colorectal carcinoma. Br J Cancer. 2001;85:1937-1 943.
Mark D. Gustavson, PhD; Annette M. Molinaro, PhD; Greg Tedeschi, BS; Robert L. Camp, MD, PhD; David L. Rimm, MD, PhD
From the Departments of Pathology (Drs Gustavson, Camp, and Rimm, and Mr Tedeschi) and Epidemiology and Public Health (Dr Molinaro), Yale University, New Haven, Conn. Dr Gustavson and Mr Tedeschi are now with HistoRx, Inc, New Haven, Conn.
Dr Gustavson and Mr Tedeschi, although not at the time of the research, are currently employed by HistoRx, Inc, a private company to which Yale University has given exclusive rights to the AQUA technology. Drs Camp and Rimm are stockholders, scientific founders, and consultants to HistoRx, Inc. Dr Molinaro has no relevant financial interest in the products or companies described in this article.
Sections of this article previously appeared in "Methods for Making Cancer Prognoses Based on the Subcellular Localization of Biomarkers," patent application No. PCT/US2007/016014, published January 17, 2008. Dr Rimm, MrTedeschi, and Drs Camp and Gustavson are listed as inventors for this patent; Yale University is listed as the assignee. To view the full patent application, access the following Web site: http://www.freepatentsonline.com/WO2008008500.html.
Reprints: Mark D. Gustavson, PhD, HistoRx, Inc, 300 George St, New Haven, CT 06511 (e-mail: firstname.lastname@example.org).
Table 1. Clinicopathologic Features of Colorectal Cancer (CRC) Cohorts * Clinicopathologic Feature Training Set Validation Set Total No. 599 447 Median survival, mo 23 (disease- 20 (recurrence-free) specific death) Median age, y <68: 301 (50.3) <70: 234 (49.1) >68: 291 (48.6) >70: 209 (43.8) Sex Female 328 (54.7) 234 (49.1) Male 264 (44.1) 208 (43.6) Histologic grade Well differentiated 184 (30.7) 190 (39.8) Moderately differentiated 230 (38.4) 126 (26.4) Poorly differentiated 60 (10.0) 31 (6.5) T pathologic stage T1 20 (3.3) 11 (2.3) T2 181 (30.2) 51 (10.7) T3 313 (52.3) 269 (56.4) T4 3 (0.5) 30 (6.3) Nodal status, No. of lymph node metastases 0 276 (46.1) 177 (37.1) 1-3 150 (25.0) 117 (24.5) [greater than or 78 (13.0) 67 (14.1) equal to] 4 * Unless otherwise indicated, data are expressed as number (percentage). The training set was constructed at the Yale Tissue Microarray (TMA) facility from 599 CRC cases obtained at Yale from 1970 through 1981. The National Cancer Institute (NCI) colon cancer TMA (validation set) was designed by statisticians at the NCI and constructed at the University of Virginia, Department of Pathology. Represented on the TMA are colon cancer specimens obtained from incident cases that occurred in members of the Kaiser Permanente Northwest Health Plan, 1989-1996. Median survival times for the training set was 23 months (disease-specific survival) and 20 months (recurrence-free survival) for the validation set. Cases are broken down by age (no information for 8 cases in training set; 32 cases in validation set), sex (no information for 7 cases in training set; 35 cases in validation set), histologic grade (no information for 125 cases in training set; 130 cases in validation set), T pathologic stage (no information for 82 cases in training set; 116 cases in validation set), and nodal status (no information for 95 cases in training set; 116 cases in validation set). Percentages are given as the percentage of the total cohort. Table 2. Multivariate Analysis--Clinical Model * Hazard Ratio, No. Variables (95% CI) (([dagger]) P Value Training set T stage ([double dagger]) 2.27 (1.56-3.29) <.001 Nodal status ([section]) 3.55 (2.38-5.39) <.001 Sex, F 0.71 (0.52-0.96) .03 Validation set T stage ([double dagger]) 2.04 (1.01-4.13) .05 Nodal status ([section]) 3.89 (2.24-6.77) <.001 Sex, F 0.89 (0.60-1.34) .59 * Cox proportional hazards multivariate analysis of clinical features that produce the best clinical model for the indicated training set (5-year disease-specific survival; n = 599) and the validation set (disease-free survival; n = 447) with indicated hazard ratios and P values. ([dagger]) CI indicates confidence interval. ([double dagger]) T3 and T4 pathologic stages. ([section]) Nodal status refers to those with [greater than or equal to] 4 lymph node metastases. Table 3. Multivariate Analysis--Testing Thymidylate Synthase (TS) Expression Ratio * Hazard Ratio, No. Variables (95% CI) ([dagger]) P Value Training set T stage ([double dagger]) 2.1 1 (1.46-3.06) <.001 Nodal status ([section]) 3.45 (2.32-5.14) <.001 TS N/C ration 1.79 (1.30-2.67) .001 Validation set T stage ([double dagger]) 2.06 (0.97-4.37) .06 Nodal status ([section]) 3.41 (1.88-6.17) <.001 TS N/C ration 1.47 (0.94-2.28) .09 * Cox multivariate proportional hazards multivariate model adding TS expression ratio to the best clinical model for the indicated training set (5-year disease-specific survival; n = 599) and the validation set (disease-free survival; n = 447) with indicated hazard ratios and P values. ([dagger]) Cl indicates confidence interval. ([double dagger]) T3 and T4 pathologic stages. ([section]) Nodal status refers to those with [greater than or equal t] 4 lymph node metastases. ([parallel]) N/C ratio indicates a nuclear to cytoplasmic ratio > 1.01.
|Gale Copyright:||Copyright 2008 Gale, Cengage Learning. All rights reserved.|