Since our original cloning of WWOX  abundant evidence has accumulated indicating that this gene likely plays a role in either tumor initiation or progression in various neoplasias including ovarian cancer [3–12]. WWOX spans the second most common chromosomal fragile site, FRA16D. This genomic region is prone to chromosome breakage, recombination and gross rearrangements. Multiple ovarian carcinoma studies have demonstrated high frequency of allelic losses for specific chromosomal regions on 16q  and since the losses have been found both in low and high-grade tumors it has been proposed to be an early event in ovarian carcinoma development . Loss of heterozygosity at 16q23.2-q24.2 has been correlated with ovarian carcinoma metastasis and advanced tumor stages . Comparative Genomic Hybridization (CGH) studies confirmed those observations also indicating high frequency of 16q21-q24 genomic losses in advanced invasive ovarian carcinoma stages [25, 26]. Taken together the data indicates that a gene, likely WWOX, located at 16q23-24 is associated with a worse prognosis in ovarian cancer.
In this study WWOX protein expression in normal ovary was predominantly observed in the OSE and epithelial inclusions, both sites proposed to be where ovarian epithelial neoplastic transformation originates. Because of this finding together with the observations of Paige (2001) , that ovarian carcinoma cell lines have high frequency of genomic losses affecting the WWOX gene including homozygous deletions, we considered it of relevance to analyze WWOX protein expression among the four major surface epithelial derived ovarian carcinoma histotypes. Our analysis by immunoblotting and IHC demonstrated that about one third of ovarian carcinomas displayed extremely reduced or absent WWOX protein expression.
It is worth noting that in all instances, normal and tumor, WWOX staining was exclusively cytoplasmic and in some serous carcinomas staining was also localized to the apical cell membrane or luminal border. This is similar to our observations in breast tumors , where WWOX was always cytoplasmic without nuclear staining. This is also in agreement with observations from other groups . This is in stark contrast with Watanabe (2003)  who observed that some cells of normal and tumor breast and gastric cases had WWOX protein expression localized to the nuclei. Possibly this discrepancy is due to the use of different antibodies.
In this ovarian cancer study, we also observed that Serous and Endometroid histotypes express normal or high WWOX but two rare but well described ovarian carcinoma histotypes, mucinous and clear cell (28,29,30), demonstrated a higher frequency of loss of WWOX protein expression. This association was more significant for the mucinous type, for which we observed that 70% (7/10) of cases were devoid of WWOX protein expression. Eventhough a small number of mucinous cases were studied the overall high frequency of decreased or loss of WWOX in these tumors as a group is intriguing. Borderline significance was found for the pure CCC for which 42% of cases demonstrated no or very low WWOX protein expression (Figure 3H versus 3G).
Primary mucinous ovarian tumors have been shown to carry a higher frequency of K-ras mutations as one of the few genetic hallmarks able to distinguish it from other histotypes, while CCC are characterized by a lack of p53 mutations. Mucinous carcinomas and CCC, has been associated with a higher frequency of resistance to chemotherapy (31). Interestingly, in our study, we observed a statistically significant trend between loss of WWOX expression and worse patient outcome. Patients whose tumors expressed WWOX at moderate or high levels fared better than those with low expression of this enzyme (Figure 4). We also found a statistically significant correlation with loss of WWOX and stage IV at initial diagnosis (FIGO). The association with overall survival could be due to an increased frequency of loss in higher stage tumors that are associated with a poorer outcome or alternatively due, in part, to an increased frequency of clear cell and mucinous tumors which have a lower likehood of responding to therapy . It was also of interest that we observed a positive correlation between WWOX and PR expression. In contrast to our previous observations in breast cancer in which we observed a strong correlation between ER and PR with WWOX expression , in this ovarian study, only PR was associated with WWOX expression (Table 2). Specifically PR negative ovarian carcinomas also lacked WWOX expression. Progestins acting through the PR have been postulated to be protective for ovarian carcinoma development [32, 33] due to a postulated ability of inhibiting cell proliferation and inducing apoptosis (34,35). We noticed that all (100%) of our analyzed mucinous carcinomas were PR negative which agrees with the literature regarding absence of PR expression in mucinous tumors [36, 37] while 90% of CCC were PR negative. These correlations raise the question of whether the observed positive association between PR loss and WWOX loss is a consequence of the predominance of WWOX loss in the two aforementioned histotypes, rather than a direct mechanistic association between WWOX and PR. Changes in progesterone signaling and regulation of progesterone-responsive genes can be of critical significance in the ovarian tumorigenic process by itself . The opposite could be true as well i.e. that high levels of WWOX expression is associated with PR expression and as a consequence related to the putative protective effect of progestins that in turn could be associated with more favorable patient outcomes.