- Top of page
Objective To investigate the role of serum inhibin A, inhibin pro-αC immunoreactivity, activin A, and follistatin in postmenopausal women with epithelial ovarian cancer.
Design Case-control study.
Sample Serum samples from 27 postmenopausal women with epithelial ovarian cancer and 54 controls from the general population participating in an ovarian cancer screening trial.
Results Women with epithelial ovarian cancer had significantly higher serum levels of pro-αC immunoreactivity (P= 0.03), activin A (P= 0.004) and follistatin (P= 0.04), but not inhibin A (P= 0.13). Using the 90th centile in the control group as the cut off, pro-αC levels were elevated in 41% of women with epithelial ovarian cancer, while inhibin A was elevated in only 15%. Using the 95th centile as the cut off, serum pro-αC was elevated in only 11% of women with epithelial ovarian cancer (3/27), while activin A was elevated in 48% (11/23). Follicle stimulating hormone levels were significantly lower in women with epithelial ovarian cancer (P= 0.01). Although, inhibin-related peptides can modulate follicle stimulating hormone levels, there was no correlation between inhibin A, pro-αC immunoreactivity, activin A or follistatin and follicle stimulating hormone.
Conclusion These data demonstrate that though there is preferential secretion of precursor forms of the α subunit rather than dimeric inhibin A by epithelial ovarian cancer, pro-αC is unlikely to be a useful tumour marker. Activin A is more commonly elevated in postmenopausal women with epithelial ovarian cancer and its role as a tumour marker in the diagnosis and screening of epithelial ovarian cancer warrants further evaluation.
- Top of page
Ovarian cancer is the fifth most common malignancy among women in the UK with 5318 cases registered in England and Wales in 19911, the majority of which are epithelial in origin. Inhibin-related peptides are useful in the differential diagnosis and surveillance of granulosa cell tumours2, but their role in epithelial ovarian cancers remains to be defined. Inhibins and activins are structurally related dimeric proteins of the transforming growth factor-β superfamily. Inhibin is a heterodimeric glycoprotein composed of a common α subunit and one of two β subunits (βA and βB), resulting in inhibin A (αβA) and inhibin B (αβB). Activin is a dimer of the two β subunits and exists as activin A (βAβA), activin B (βBβB) and activin AB (βAβB). Follistatin is a structurally distinct, single chain glycoprotein, which is the major binding protein for activin and inhibin.
Using a nonselective assay, Healy et al.3 and Blaakaer et al.4 found that 25%–60% of women with epithelial ovarian cancer had elevated serum inhibin levels. More recently, using specific immunoassays measuring inhibin A, elevated levels were found in 5–31% of women with epithelial ovarian cancer5–7. Serum inhibin B was detected in these women, but levels were within the normal range5,8. The serum also contained immunoreactive forms of the α subunit which were not linked to the β subunit. Some of these precursor forms of inhibin (e.g. pro-αC and pro-αN-αC) could be measured using the pro-αC assay9. It has been stated that among the inhibin assays, measurement of pro-αC immunoreactivity may be of most value in women with epithelial ovarian cancer2,5. Welt et al.10 found that activin A and follistatin were secreted by epithelial ovarian cancer in vitro, and that serum activin A levels were significantly raised in women with epithelial ovarian cancer compared with controls. No studies have looked at serum follistatin levels in these women.
The present study was undertaken to investigate the role of serum inhibin A, pro-αC, activin A and follistatin in postmenopausal women with epithelial ovarian cancer.
- Top of page
The epithelial ovarian carcinomas included in the study were mucinous (n= 4), serous (n= 10), endometroid/clear cell (n= 4), undifferentiated (n= 5), transitional cell (n= 1) and borderline (n= 3) (Table 1). Five patients were Stage I, one Stage II, 14 Stage III and seven Stage IV according to the International Federation of Gynaecology and Obstetrics (FIGO) criteria. The median age of the cases was 68 years (range 47–85) and that of the controls was 67.5 years (range 51–80).
Table 1. Results of the assays for women with epithelial ovarian cancer. Values are given as n. NS = not sufficient serum sample for assay; LH = luteinising hormone; FSH = follicle stimulating hormone.
|Histology||Case no.||Age||Stage||LH (U/L)||FSH (U/L)||Pro-αC (pg/mL)||Inhibin A (pg/mL)||Activin A(pg/mL)||Follistatin (pg/mL)|
There was no significant difference between inhibin A levels in women with epithelial ovarian cancer and controls (P= 0.13). In contrast, women with epithelial ovarian cancer had significantly higher serum levels of inhibin forms containing pro-αC immunoreactivity (P= 0.03), activin A (P= 0.004) and follistatin (P= 0.04) than healthy controls (Fig. 1). An analysis of the sensitivity of the markers for epithelial ovarian cancer at cut off points defined by the 90th and 95th centiles in the control group is summarised in Table 2. The most sensitive single marker was activin A which achieved a sensitivity of 48%. Follicle stimulating hormone levels were significantly lower in women with epithelial ovarian cancer (P= 0.01) while there was no significant difference in the luteinising hormone levels (P= 0.2). There was no correlation between serum inhibin A, pro-αC immunoreactivity, activin A, follistatin or the ratio of inhibin A/activin A or follistatin/activin A and serum follicle stimulating hormone. As anticipated, follicle stimulating hormone correlated positively with luteinising hormone (rs= 0.74) and negatively with age (rs=−0.5) in women with ovarian cancer.
Figure 1. Serum concentrations (pg/mL) of inhibin A, pro-αC immunoreactivity, activin A, follistatin, FSH and LH for postmenopausal women with epithelial ovarian cancer and healthy age matched postmenopausal controls (N) from the general population. The concentrations are plotted on a logarithmic scale. Median values are highlighted. Undetectable levels have been given the numerical sensitivity limit. 38/54 controls (70%) and 15/27 women (55%) with epithelial ovarian cancer had undetectable inhibin A levels. *There was significant difference between women with epithelial ovarian cancer and controls.
Download figure to PowerPoint
Table 2. Sensitivity of the markers for epithelial ovarian cancer at cut offs defined by the 90th and 95th centiles in the control group. Values are given as %(nelevated/ntotal).
| ||90th centile of controls (pg/mL)||Sensitivity using 90th centile as cut off||95th centile of controls (pg/mL)||Sensitivity using 95th centile as cut off|
|Inhibin A||37||14.8 (4/27)||39||14.8 (4/27)|
|Pro-αC||201||40.7 (11/27)||582||11.1 (3/27)|
|Activin A||1082||52.2 (12/23)||1163||47.8 (11/23)|
|Follistatin||940||11.1 (3/27)||1337||0 (0/27)|
|Pro-αC + Activin A||–||70.4 (19/27)||–||51.9 (14/27)|
A number of other observations in the study were noteworthy (Table 1). First, the few patients with inhibin A elevations had serous (n= 3) and endometroid (n= 1) cancers, while low or undetectable levels were documented in the five mucinous carcinomas (including the borderline mucinous tumour). Second, activin A was elevated in four of five undifferentiated tumours and all three endometroid cancers. Finally, surprisingly high levels of pro-αC were observed in two postmenopausal women in the control group who have remained apparently healthy in the year they have been followed up as part of the screening trial.
Our data revealed no significant difference between serum inhibin A concentrations in women with epithelial ovarian cancer and age matched controls, with levels elevated (> 95th centile for controls) in only 15% of cases. Other recent studies, which measured dimeric inhibin A, revealed similar elevations in 5–13% of women with epithelial ovarian cancer5,6,22. None of the five mucinous tumours in our study were associated with raised levels. This is in contrast to Burger et al.6 and Robertson et al.22, who noted elevations in three of 12 and four of 20 mucinous carcinomas, respectively. The differences may reflect the small numbers of mucinous tumours in most of these series. Overall the consistent message is that dimeric inhibin A is less informative than other inhibin-related peptides in epithelial ovarian cancer. Frias et al.23 have recently reported that pre-operative serum inhibin A may be an independent prognostic factor for survival in postmenopausal women with epithelial ovarian cancer.
The absolute concentrations of pro-αC containing inhibin forms were approximately tenfold higher than dimeric inhibin A levels in the healthy postmenopausal controls, confirming previous findings9. Two healthy postmenopausal women had very high levels of pro-αC immunoreactivity, for which no cause has yet been established. In women with epithelial ovarian cancer, using the 90th centile of controls as the cut off, serum pro-αC immunoreactivity was found to be elevated in 41%, while inhibin A levels were elevated in only 15%. This supports the findings of Lambert-Messerlian et al.5 and Burger et al.2 that there is preferential secretion of precursor forms of the α subunit, rather than dimeric inhibin A, by epithelial ovarian cancer. However, serum pro-αC is unlikely to be a useful tumour marker in epithelial ovarian cancer as levels were elevated in only 11% of women with epithelial ovarian cancer, using the more stringent 95th centile as cut off. A recently published large study of inhibin forms in postmenopausal women with epithelial ovarian cancers concluded that assays detecting all inhibin forms containing the α subunit, and not just those detecting the pro-αC subunit, will probably prove to the most useful screening test22. The newly developed αC immunofluorometric assay, which detects all α subunit-containing proteins, has been shown to have increased sensitivity for detection of ovarian cancers, especially in combination with CA12524.
The most commonly elevated marker was activin A, with undifferentiated and endometroid tumours accounting for the majority of the elevations. Serum activin levels were elevated in 52% of women with epithelial ovarian cancer. Lambert-Messerlian et al.25 found 63% of women with epithelial ovarian cancer had an elevated pre-operative level, using the 90th centile of controls as the cut off. There is increasing evidence both from immunohistochemical analysis and expression studies on cancer cell lines that there is differential production and expression of inhibin/activin subunits with increased production of the inhibin β but not the α subunit in epithelial ovarian cancer26–28. Activin A is secreted more frequently than inhibin in tumour explants10. Activin receptors have been found in epithelial ovarian cancer cell lines10,26 and autocrine/paracrine interactions of activin and follistatin are probably involved in signal modulation and tumour cell proliferation26,29. The significance of elevated activin levels remains to be elucidated. Although preliminary data from Petraglia et al.8 and Frias et al.23 found no correlation with clinical disease, a recent study22, using serial assays in women with epithelial ovarian cancer, has found activin A to be a marker of persistent and recurrent disease.
Total serum follistatin was significantly elevated in women with epithelial ovarian cancer, but this was of little clinical value as a marker. No individual women had a serum follistatin concentration that exceeded the normal range. Only one previous study has examined serum follistatin levels in cancer. Elevated serum free follistatin were found in a group of 39 women with solid cancers that included four ovarian malignancies14. Ovarian cancer cell lines have been found to express follistatin mRNA and to secrete follistatin26,30. However, there is growing evidence that serum follistatin may not be primarily an ovarian product17,31. Thus, even if cancers produce follistatin, it may not be reflected in serum measurements. In addition, follistatin is present in a number of isoforms in biological fluids. Recent work suggests that follistatin 315 is the dominant variant in serum32. There are no standard assays for measuring follistatin and the lower sensitivity of the current assay for follistatin 315 compared with follistatin 288 may result in an underestimate of the total amount of follistatin17.
Serum follicle stimulating hormone levels have been found to be significantly lower in women with epithelial ovarian cancer compared with controls33,34. Inhibin-related peptides can modulate follicle stimulating hormone levels and an earlier report had shown an inverse correlation between immunoreactive inhibin and follicle stimulating hormone in epithelial ovarian cancer35. However, this was not borne out in any of the larger series3,22. Our results further confirmed the lack of correlation between dimeric inhibin A, pro-αC, activin A or follistatin and follicle stimulating hormone levels in women with epithelial ovarian cancer.
In summary, the data demonstrates that though there is preferential secretion of precursor forms of the α subunit rather than dimeric inhibin A by epithelial ovarian cancer, pro-αC is unlikely to be a useful tumour marker in postmenopausal women with epithelial ovarian cancer. Activin A is more commonly elevated and its role as a serum tumour marker in the diagnosis and screening of epithelial ovarian cancer warrants further evaluation.
The authors would like to thank Ms C. Balfour for performing the enzyme linked immunosorbent assays. Dr U. Menon was funded by a Clinical Training Fellowship Grant from the St Bartholomew's Hospital Joint Research Board. Dr C. Bose was funded by the Indian National Science Academy/Royal Society Exchange Fellowship.