| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Oncology, Markers, Clinical Correlates |
Biomedical Sciences Program [G. S.], the Department of Medicine [S. B. H., N. M. V.], and the Cancer Center, University of California, San Diego, La Jolla, California 92093; Department of Gynecology, University of Zurich, Zurich, Switzerland 8091 [D. F.]; Gynecology Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY 10021 [A. H., W. J. H.]; and Arizona Cancer Center, University of Arizona, Tucson, Arizona 85721 [D. S. A.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Several studies have suggested a relationship between MMR status and sensitivity to certain classes of chemotherapeutic agents (3, 4, 5, 6) . MMR-deficient cells are resistant to the platinum-containing drugs cisplatin and carboplatin, both of which produce similar adducts in DNA and are particularly important because of their widespread clinical use. The MMR-deficient HCT116 colon cancer and HEC59 endometrial cancer cell lines both exhibit low-level resistance to cisplatin compared to their MMR-proficient counterparts (4) . In addition, an MSH2-knockout murine embryonic stem cell line was found to be 2.1-fold more resistant to cisplatin than its isogenic MMR-proficient parental counterpart (3) .
Further evidence linking loss of MMR function to resistance to cisplatin is provided by the observation that selection for cisplatin resistance in vitro is sometimes accompanied by loss of expression of some MMR proteins. A cisplatin-resistant derivative of the human ovarian adenocarcinoma cell line 2008 was found to have lost hMLH1 protein expression (4) . In another study, 9 of 10 sublines of the drug-sensitive A2780 ovarian cancer cell line that were independently selected for acquired cisplatin resistance were found to be deficient in hMLH1. Western blot analysis of human ovarian carcinoma samples obtained directly from patients before initial chemotherapy demonstrated that 10% lacked MLH1 expression, whereas in tumor samples obtained from other patients after platinum-based chemotherapy, 36% lacked MLH1 expression (5) . Finally, a limited immunohistochemical study of paired ovarian cancer samples obtained from the same patient before and after at least three cycles of platinum-based primary chemotherapy demonstrated that MLH1 expression was reduced in 66% of cases (6) .
The increased mutation rate that occurs throughout the genome in cells that lack MMR is often detected as microsatellite instability (7) . An estimated 15–20% of sporadic ovarian tumors exhibit microsatellite instability (5) , and there is a high probability that an even larger fraction contain at least some cells that are MMR-deficient. Presently available evidence suggests that resistance should emerge more rapidly in tumors that contain a larger number of MMR-deficient cells (1) , and this may be the basis for the acquired resistance to primary chemotherapy that emerges so frequently in patients with ovarian cancer.
In this study, we used immunohistochemical staining techniques to quantify hMLH1 and hMSH2 expression in paired ovarian tumor sections from patients before and after platinum-based treatment. All clinical data on these patients had been collected prospectively as they had all been treated as part of either a Memorial Sloan-Kettering Cancer Center or Southwest Oncology Group trials (8) . We sought associations between expression of these two MMR proteins and parameters known to be of prognostic significance as well as response to treatment and overall survival.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
.
|
Immunohistochemical Analysis.
Each slide was scored blindly for the percentage of malignant cells
that stained positively (0–100% in 10% increments) and intensity of
staining (0–4) for hMLH1 or hMSH2 by an independent
pathologist who was unaware of the treatment status of the patient from
whom the tissue sample was obtained.
Statistical Analysis.
Based on available clinical and staining data, the following
associations were examined for patients before treatment, after primary
treatment, and for the difference between pre- and posttreatment:
percent staining versus intensity of staining of hMLH1 and
hMSH2 (Spearman correlation coefficients); percent and intensity of
staining for hMLH1 versus hMSH2 (Spearman correlation
coefficients); and percent and intensity of staining of hMLH1 and hMSH2
versus grade, stage, histological subtype (Kruskall-Wallis
ANOVA), CA125 levels (for patients 1–41 only), age, response to
treatment (Spearman correlation coefficients), and overall survival
(comparison of Kaplan-Meier curves by log-rank statistic). In addition,
patients were grouped based on their change in percent staining for
hMLH1 and hMSH2 after primary platinum-based therapy and analyzed for
association with overall survival (comparison of Kaplan-Meier curves by
log-rank statistic).
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Fig. 1
shows the
distribution of the time elapsed between the pretreatment and
subsequent biopsy for all 54 patients. The median time between the
pretreatment and posttreatment tumor samples for all patients was 6
months (range, 2–63 months). The median number of cycles was 4 (range,
2–10). Fig. 2
shows the distribution of
the percent of malignant cells stained in the samples obtained before
and after treatment. The mean change in percent staining of cells for
hMLH1 (pre- minus posttreatment) was 18.7 ± 5.7% (SE). In the
case of staining for hMSH2, mean change in percent staining of cells
was 15.0 ± 5.1% (SE). For both hMLH1 and hMSH2, the difference
in percent staining after platinum-based treatment was found to be
statistically significant using sign-rank analysis (P =
0.0020 for hMLH1 and 0.0179 for hMSH2). Thus, the expression of both
proteins tended to be lower in the posttreatment samples. Subset
analyses were undertaken to determine whether the magnitude of the
decrease in staining varied as a function of the time elapsed between
the first and second biopsies. Even in patients whose second biopsy was
obtained 
4 months after the first, a trend toward decreased staining
was observed for both hMLH1 and hMSH2, although due to the small number
of patients in this subset, the changes did not reach statistical
significance. The magnitude of the decrease was not significantly
greater than the mean for all patients in the subset whose second
biopsy was 
10 months after the first.
|
|
shows that there were highly
significant correlations between the fraction of the malignant cells in
a tumor sample that stained positive for hMLH1 and the intensity of the
staining for hMLH1 in the same tissue sample. This was also observed
for hMSH2. Strong correlations were observed in both the pretreatment
and posttreatment tumor samples. In addition, the difference in percent
of cells staining positively between the pre- and posttreatment samples
within the same patient was correlated with the difference in intensity
of staining between the pre- and posttreatment samples. Thus, tumors in
which a large fraction of cells stained positively also tended to stain
with greater intensity, suggesting that these two staining
characteristics are not independent of each other.
|
.
Likewise, the difference in percent of cells staining positively for
hMLH1 between the pre- and posttreatment samples within the same
patient was correlated with the difference in the percent of cells
staining positively for hMSH2. Similar correlations were found between
the intensity of staining for hMLH1 and the intensity of staining for
hMSH2 in the pre- and posttreatment samples, and the differences in
intensity of staining for the two proteins in the pre- and
posttreatment samples. Thus, tumors that expressed one of the proteins
at a high level tended to express the other at a high level and
vice versa. When there was a change in expression of one
protein between the pre- and posttreatment samples from the same
patient, there tended to be a change in the other protein in the same
direction. These results provide evidence for coordinate regulation of
the level of hMLH1 and hMSH2 in ovarian cancer.
|
, posttreatment staining
for MLH1was greatest in patients in the progressive disease category
and least for patients who eventually attained a complete response
(r = -0.35995; P = 0.0081). As
expected, intensity of staining of hMLH1 in the same group of patients
also correlated with response to treatment (r =
-0.32447; P = 0.0178). Interestingly, no association
was found between the hMLH1 staining characteristics of the
pretreatment samples and response, nor between the difference in
staining characteristics between the pre- and posttreatment samples and
response. These results suggest that, contrary to previous findings,
lower expression of hMLH1 does not in itself indicate a lack of
in vivo sensitivity to platinum-based therapy. No
association was found between hMSH2 staining (percent of cells positive
or intensity) and response to treatment for either the pre- or
posttreatment samples.
|
With respect to FIGO stage, no associations were found for hMLH1, but
there were statistically significant differences in pretreatment
intensity (P = 0.0236), and in the change in intensity
(P = 0.0270), in the case of hMSH2. As shown in Fig. 4
, patients with higher stage had a
significantly lower median staining intensity.
|
shows that the percent of
malignant cells staining positive before treatment was associated with
histological subtype for both hMLH1 (P = 0.0156) and
hMSH2 (P = 0.0218). As expected from the finding that
percent staining correlated well with intensity of staining, the
intensity of staining of hMLH1 and hMSH2 in patients before treatment
was also associated with histological subtype (P =
0.0253 and 0.0096, respectively), with a majority of patients with
serous histology staining for hMLH1 and hMSH2 at an intensity of 3 and
4 before treatment.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The first observation of interest is that staining for both hMLH1 and hMSH2 was demonstrable in a large fraction of the malignant cells in most human ovarian carcinomas before the start of treatment. Second, there was a statistically significant difference in the expression of both proteins and in the percentage of malignant cells expressing these proteins after platinum-based therapy. Subset analyses did not detect any clear difference in the magnitude of the decrease as a function of the timing of the second biopsy, although the small size of the subsets limited the power of this analysis. Thus, after drug exposure, the persistent tumor mass was biologically altered with respect to the expression of these two proteins. This observation is consistent with a previous smaller study, which disclosed that hMLH1 expression was decreased in 66% of 38 patients with ovarian carcinoma after at least three cycles of primary chemotherapy (6) . These results are in contrast to the initial report of another study by Mackean et al. (9) , which examined the prognostic significance of MMR deficiency. Immunohistochemical staining of paraffin-embedded samples in the study by Mackean et al. (9) did not identify any significant change in hMLH1 or hMSH2 expression in 26 paired ovarian carcinoma samples after cisplatin-based chemotherapy. There has been speculation that loss of hMLH1 expression in vitro may be due to hypermethylation of its promoter (10) , although this mechanism has yet to be studied in detail in vivo. There has been no investigation, to our knowledge, into the mechanism of loss of hMSH2 expression in these tumors.
The strongest and most consistent association identified in the present study was that between the fraction of the malignant cells that stained positive for either hMLH1 or hMSH2 and the intensity of staining for that protein in the same sample. Highly significant correlations were found for both hMLH1 and hMSH2 in both pre- and posttreatment samples. A priori it is difficult to know which of these two parameters might be more important. There is a reasonable biological basis for anticipating that they might be independent variables. For example, a few cells that have lost all hMLH1 or hMSH2 expression may be more important to the development of drug resistance than a large number of cells that have low intensity staining. However, the results of this study indicate that, in the case of human ovarian carcinoma, these two variables are clearly associated. The larger the fraction of tumor cells that stained positively, the higher the intensity of staining for that sample. This suggests that it is reasonable to search for associations between either one of these and clinical parameters separately rather than trying to develop a scoring system based on contributions from each as has been reported previously (9) .
One interesting but unexpected association identified was that between the staining characteristics of hMLH1 and those of hMSH2. Both the percent of malignant cells that stained positive and the intensity of staining for hMLH1 were well correlated with the same parameters for hMSH2. This association was demonstrated for the pretreatment and posttreatment samples and for the difference following treatment. Previous studies have reported that loss of hMLH1 protein expression is coupled with loss of PMS2 protein expression (5) . The loss of PMS2 was ascribed to the lack stabilization by its heterodimeric partner hMLH1. Our results are consistent with a similar coordinate regulation of hMLH1 and hMSH2 levels in ovarian cancer.
Based on the observation that loss of MSH2 is associated with cisplatin and carboplatin resistance in vitro (3) and that loss of hMLH1 expression frequently accompanies acquired cisplatin resistance in vitro (5) , one would expect that low hMLH1 or hMSH2 expression would correlate with a poor response to these agents in vivo. However, our results indicate only a relatively weak association between staining characteristics and therapeutic outcome, and this association was in the opposite direction than that predicted by the in vitro data. In the case of hMLH1, response did differ as a function of the fraction of cells staining positively for hMHL1, and this association was corroborated by the fact that the same association was observed for hMLH1 staining intensity. However, patients attaining a complete response had a lower mean intensity of staining of hMLH1 posttreatment, compared to other patients (P = 0.0178), and patients who had disease progression had a higher mean percent staining of hMLH1 posttreatment, compared with other patients (P = 0.0081). In addition, there was no association of hMSH2 staining with response, and no association of staining for either protein with overall survival. Thus, these results do not support the hypothesis that lack of hMLH1 or hMSH2 staining predicts in vivo resistance to the platinum drugs.
Associations were identified between pretreatment CA125 levels and pre- versus posttreatment difference in staining intensity for hMLH1 and hMSH2. Even more interesting, an association was identified between the change in CA125 level after treatment and pretreatment intensity as well as the change in intensity of staining of hMSH2. Higher CA125 levels before treatment correlated with a greater fraction of malignant cells staining positive for both proteins. A decrease in CA125 level after treatment correlated with a higher pretreatment intensity of staining as well as a decrease in intensity of staining of hMSH2. This is an important observation in that changes in CA125 levels have been used as an indirect measure of tumor responsiveness, with a decrease in CA125 levels indicating a better response to treatment (11) .
A relation was found between tumor stage and hMSH2 staining. With respect to both the fraction of positively staining cells and intensity of staining, the data permit rejection of the null hypothesis that all stages expressed hMSH2 at the same level. Patients with stage 4 tumors had significantly lower hMSH2 staining intensity in pretreatment samples compared with stage 3 patients (P = 0.0236). Therefore, patients with a worse prognosis based on FIGO stage had decreased hMSH2 expression before treatment. This is in contrast to a previous study in which MMR deficiency, as assessed by microsatellite instability, was demonstrated in 75% of stage 1 patients with ovarian neoplasms compared to 11% of stage 2, 3, or 4 patients (12) .
The next most interesting observation was that tumors of different histological subtypes had different levels of expression of both hMLH1 and hMSH2. Serous adenocarcinomas stood out as having greater staining (both in terms of the fraction of cells that stained positively and the intensity of staining) than the other histological subtypes. It is of interest that serous carcinoma are also generally the most responsive to platinum drug-based chemotherapy, but the data do not permit the conclusion that there is a causal link. The analysis demonstrated that there was no discernible association between MMR protein expression and FIGO grade. This was not surprising because there was only a limited distribution of grades represented, with almost all patients having grade 2 or 3 tumors.
Our results also indicate that there was no association between MMR protein expression and overall survival. Based on previous findings that MMR deficiency is associated with acquired resistance to cisplatin treatment in vitro, it was expected that expression of MMR proteins be predictive of survival. However, our analysis did not discern a correlation between either hMLH1 or hMSH2 expression and overall survival. It would have been of interest to examine for an association with progression-free survival, but information on the time to progression of disease was not available for the patients in the present study.
The association between loss of MMR activity and resistance to cisplatin and carboplatin has now been very well established using molecularly engineered tumor cell lines. However, it is not clear that immunohistochemical staining for hMLH1 or hMSH2 measures MMR activity even indirectly. Unfortunately, tools for making direct measurements of MMR activity in human tumor samples are not available, and thus it was important to explore the question of whether immunohistochemical staining for hMLH1 or hMSH2 could be used as a surrogate. This study contributes importantly to the field by suggesting a negative answer to this question.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by Grant CA 78648 and Southwest
Oncology Group Grants CA 38926 and CA 32102 from the National
Institutes of Health and by the Swiss National Sciences Foundation
(31-52531.97). This work was conducted in part by the Clayton
Foundation for Research, California Division. Dr. Howell is a Clayton
Foundation investigator. This study was performed as an Institutional
Review Board-approved protocol. 
2 To whom requests for reprints should be
addressed, at Department of Medicine 0058, University of California,
San Diego, La Jolla, CA 92093. Phone: (858) 822-1110; Fax: (858)
822-1111; E-mail: gsamimi{at}ucsd.edu
3 The abbreviations used are: MMR, DNA mismatch
repair; FIGO, Fédération Internationale de
Gynécologie et d’obstetrique.
Received 9/23/99; revised 12/28/99; accepted 12/29/99.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Chen, X. He, Z. Wang, D. Wu, H. Zhang, C. Xu, H. He, L. Cui, D. Ba, and W. He Identification of Human T Cell Receptor {gamma}{delta}-recognized Epitopes/Proteins via CDR3{delta} Peptide-based Immunobiochemical Strategy J. Biol. Chem., May 2, 2008; 283(18): 12528 - 12537. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. K. Lee, R. T.P. Poon, J. Y. Wo, S. Ma, X.-Y. Guan, J. N. Myers, P. Altevogt, and A. P.W. Yuen Lupeol Suppresses Cisplatin-Induced Nuclear Factor-{kappa}B Activation in Head and Neck Squamous Cell Carcinoma and Inhibits Local Invasion and Nodal Metastasis in an Orthotopic Nude Mouse Model Cancer Res., September 15, 2007; 67(18): 8800 - 8809. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Flanagan, B. W. Robinson, C. M. Krokosky, and D. S. Shewach Mismatched nucleotides as the lesions responsible for radiosensitization with gemcitabine: a new paradigm for antimetabolite radiosensitizers Mol. Cancer Ther., June 1, 2007; 6(6): 1858 - 1868. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Lin and S. B. Howell DNA mismatch repair and p53 function are major determinants of the rate of development of cisplatin resistance Mol. Cancer Ther., May 1, 2006; 5(5): 1239 - 1247. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Chirnomas, T. Taniguchi, M. de la Vega, A. P. Vaidya, M. Vasserman, A.-R. Hartman, R. Kennedy, R. Foster, J. Mahoney, M. V. Seiden, et al. Chemosensitization to cisplatin by inhibitors of the Fanconi anemia/BRCA pathway. Mol. Cancer Ther., April 1, 2006; 5(4): 952 - 961. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. O'Brien and R. Brown Signalling cell cycle arrest and cell death through the MMR System Carcinogenesis, April 1, 2006; 27(4): 682 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Viens, T. Petit, A. Yovine, P. Bougnoux, G. Deplanque, P.-H. Cottu, R. Delva, J.-P. Lotz, S. V. Belle, J.-M. Extra, et al. A phase II study of a paclitaxel and oxaliplatin combination in platinum-sensitive recurrent advanced ovarian cancer patients Ann. Onc., March 1, 2006; 17(3): 429 - 436. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Fukushima, Y. Katayama, T. Watanabe, A. Yoshino, A. Ogino, T. Ohta, and C. Komine Promoter Hypermethylation of Mismatch Repair Gene hMLH1 Predicts the Clinical Response of Malignant Astrocytomas to Nitrosourea Clin. Cancer Res., February 15, 2005; 11(4): 1539 - 1544. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Di Nicolantonio, L. A. Knight, P. A. Whitehouse, S. J. Mercer, S. Sharma, P. A. Charlton, D. Norris, and I. A. Cree The ex vivo characterization of XR5944 (MLN944) against a panel of human clinical tumor samples Mol. Cancer Ther., December 1, 2004; 3(12): 1631 - 1637. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Gifford, J. Paul, P. A. Vasey, S. B. Kaye, and R. Brown The Acquisition of hMLH1 Methylation in Plasma DNA after Chemotherapy Predicts Poor Survival for Ovarian Cancer Patients Clin. Cancer Res., July 1, 2004; 10(13): 4420 - 4426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Samimi, N. M. Varki, S. Wilczynski, R. Safaei, D. S. Alberts, and S. B. Howell Increase in Expression of the Copper Transporter ATP7A during Platinum Drug-Based Treatment Is Associated with Poor Survival in Ovarian Cancer Patients Clin. Cancer Res., December 1, 2003; 9(16): 5853 - 5859. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Kishi, Y. Doki, M. Yano, T. Yasuda, Y. Fujiwara, S. Takiguchi, S. Kim, I. Higuchi, and M. Monden Reduced MLH1 Expression after Chemotherapy Is an Indicator for Poor Prognosis in Esophageal Cancers Clin. Cancer Res., October 1, 2003; 9(12): 4368 - 4375. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Rau, I. Sturm, H. Lage, S. Berger, U. Schneider, S. Hauptmann, P. Wust, H. Riess, P. M. Schlag, B. Dorken, et al. Dynamic Expression Profile of p21WAF1/CIP1 and Ki-67 Predicts Survival in Rectal Carcinoma Treated With Preoperative Radiochemotherapy J. Clin. Oncol., September 15, 2003; 21(18): 3391 - 3401. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. N. Manderson, A.-M. Mes-Masson, J. Novak, P. D. Lee, D. Provencher, T. J. Hudson, and P. N. Tonin Expression Profiles of 290 ESTs Mapped to Chromosome 3 in Human Epithelial Ovarian Cancer Cell Lines Using DNA Expression Oligonucleotide Microarrays Genome Res., January 1, 2002; 12(1): 112 - 121. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Karran Mechanisms of tolerance to DNA damaging therapeutic drugs Carcinogenesis, December 1, 2001; 22(12): 1931 - 1937. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Niedner, R. Christen, X. Lin, A. Kondo, and S. B. Howell Identification of Genes That Mediate Sensitivity to Cisplatin Mol. Pharmacol., December 1, 2001; 60(6): 1153 - 1160. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. K. Sood, R. Holmes, M. J. C. Hendrix, and R. E. Buller Application of the National Cancer Institute International Criteria for Determination of Microsatellite Instability in Ovarian Cancer Cancer Res., June 1, 2001; 61(11): 4371 - 4374. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
