Abnormally-fucosylated haptoglobin: a cancer marker for tumour burden but not gross liver metastasis.

A previous study has shown that there are high levels of an abnormally-fucosylated form of haptoglobin (FHp) in the blood of cancer patients (Thompson & Turner, 1987b). In this study, we investigated the expression of this substance in serial blood specimens from women with ovarian or breast cancer who were undergoing cytotoxic chemotherapy. The level of FHp was related to patient response to therapy status, this latter index being an indirect determination of tumour burden. FHp levels did not correlate with gross liver metastasis (as shown by CT scans or the blood levels of liver enzymes). This conclusion was further supported by results from patients with hepatocellular cancer. FHp was elevated in most of these patients, but the pattern of change did not correlate with variations in the level of the hepatoma marker, alpha-foetoprotein. It seems likely that FHp is produced by the liver. Primary and secondary tumours could release substances, such as cytokines, which interfere with fucose metabolism in the liver. ImagesFigure 1Figure 2Figure 3Figure 4

We have previously reported that cancer sera have high levels of an abnormal form of haptoglobin (FHp) that can be extracted from blood using the fucose-specific lectin Lotus tetragonolobus (Thompson & Turner, 1987a). The expression of this molecule is related to the amount of protein-bound fucose in the blood, but is not related to the total haptoglobin concentration (Thompson & Turner, 1987b). The same molecule was not present in the blood from healthy individuals or in individuals with inflammatory diseases such as rheumatoid arthritis (Thompson & Turner, 1987b) or bronchial pneumonia (Thompson & Turner, 1988) despite the haptoglobin levels in these diseases being frequently elevated. Additional studies of 11 serial specimens from two ovarian cancer patients showed that elevated levels of FHp were associated with increasing tumour burden (Thompson et al.,1987b). In order to confirm and extend these observations we have further investigated FHp by measuring its level in 150 serial specimens from 21 women with ovarian cancer and 27 women with breast cancer. As the cause of the changes in the fucosylation of haptoglobin is unknown, we have explored the possibility that elevated FHp levels result from the presence of metastatic deposits in the liver.

Materials and methods
Patients Blood specimens were obtained by venepuncture from 21 women with ovarian cancer (median age = 60 yr; range 33-79); from 27 women with breast cancer (median age = 60 yr; range 40-77); and from nine women and seven men with hepatocellular cancer. Sera were separated by low speed centrifugation for 10 min and were stored at -20°C. All analyses were carried out within 2 years of collection. Ovarian cancers were diagnosed by laparotomy (stages III/IV) and confirmed by histology (serous or mucinous adenocarcinoma). At laparotomy, different amounts of tumour were removed; in all cases tumour remained in the abdomen, despite attempts at total surgical debulking, but for some patients this was a very small amount. During the period of specimen collection, 17 patients received several courses of Carboplatin chemotherapy, one patient received several courses of Cisplatin, one patient received several courses of Chlorambucil, one patient received several courses of Mitoxanthrone and another patient received several courses of Treopsulphan. Some patients had had therapy with Cisplatin prior to receiving Carboplatin.
The 27 women who had breast cancer had undergone either a lumpectomy or a mastectomy 6 months to 20 years previously (median 3.5 yr). Twenty-four women still had clinical signs of cancer when specimen collection was started. During the period of collection the major therapies were as follows: Mitoxanthrone (ten women); aminoglutethimide, hydrocortisone, disodium pamidronate (three); aminoglutethimide, hydrocortisone (three); Tamoxifen (two); Cyclophosphamide, 5-fluoruracil, prednisolone (two); Ifosphamide, Doxorubicin (one); Vincristine, prednisolone (one); Vincristine, mitomycin (one); LHRH agonist (one); radiotherapy (one) and two were untreated. Nineteen of the women were previously treated with radiotherapy; 17 of these received Tamoxifen and five had prior treatment with Mitoxanthrone.
At least two blood specimens were collected from each ovarian and breast cancer patient (median number three; range two to six) over a period of time that varied (median time = 12 months; range 1-22 months). The time between collections also varied (median = 3 months; range 1-15 months). Seventy-seven and 73 specimens were collected from the ovarian and breast patients respectively. Only one or two specimens were collected from the hepatoma patients.
Ovarian and breast cancer patients were assessed when specimens were collected; these assessments were made by clinical examination and appropriate radiological and ultrasound scanning techniques. The patient response to therapy was categorised into three groups; complete response, partial response or stable disease, and progressive disease. Complete response was defined as complete disappearance of all demonstrable disease for at least 4 weeks. Partial response was defined as a reduction (> 50% bidimensional or > 30% unidimensional) in total size of measurable disease, and stable disease was defined as < 50% change (bidimensional) or < 30% (unidimensional) in the total size of the measurable lesions. Progressive disease was defined as a >50% (bidimensional) or >30% (unidimensional) increase in the size of any measurable lesion.
The hepatocellular cancers were diagnosed by physical examination combined with radiological and ultrasonic scanning techniques, and confirmed by the histological examination of biopsy material. No information was available for the clinical assessments of the hepatoma patients. Identification of Serum FHp These methods have been extensively described elsewhere (Thompson & Turner, 1987a,b;Thompson et al., 1987). Briefly, a fucose-binding lectin (Lotus tetragonolobus, Sigma) was coupled to CNBr-activated Sepharose beads (Pharmacia) at 2mgml-' beads. Fucoproteins were extracted from 751 l aliquots of serum by mixing with 75 jil lotus lectin-beads for 1.5 h at 25°C. Unbound proteins were removed by six washes 0.05 mol 1-' Tris-HCl, pH 7.4, containing 25 mmol 1' KCl, 5 mmol -1 CaCl2, 5 mmol 1' MgCl2 and 0.5% (v/v) Nonidet P40. In preliminary experiments, bound fucoproteins were eluted from the beads either by incubating for 30 min with 50 pl of the Tris washing buffer containing 1.0 mol 1' fucose or by solubilising in 50 jil of 125 mmol 1-i Tris-HCI, pH 6.8, containing 0.35 mol l-1' sodium dodecyl sulphate (SDS), 2.7 mol l-' glycerol, 1 mmol l' EDTA and 2.9 mmol lbromophenol blue. Eluted material was reduced by boiling for 5 min in 5% (v/v) P-mercaptoethanol, and the denatured fucoproteins (12 Al extract) were separated in 8% (w/v) polyacrylamide slab gels using the discontinuous Laemmli buffer system. A silver staining procedure (Thompson, 1987) was used to visualise the separated proteins and the FHp band was identified by its molecular weight. For some specimens, the identity of this component was further confirmed by Western blotting (Thompson & Turner, 1987b). Any specimen that was haemolysed was not analysed for FHp. Comparisons of the electrophorectic patterns of serum fucoproteins extracted by either fucose of SDS are given in Figure 1. Patterns are also shown for sera that were extracted with Sepharose beads alone. It can be seen that both fucose and SDS extract a molecule that migrates at 40-45 kDaltons (subsequently identified as the P-chain of haptoglobin by Western blotting). This band is only seen if the specimen is extracted with Sepharose beads coupled to lotus lectin. As the SDS procedure usually removed larger amounts of the 40-45 kDalton component, this method was routinely employed for extraction of fucosylated haptoglobin from the cancer sera. Silver-stained electrophoretic separations of lotus-extracted sera from eight women with ovarian cancer. All women were undergoing cytotoxic chemotherapy. Groups 1-3 are from women who showed a complete response to therapy; groups 4-6 are from women who had stable disease followed by progressive disease; and groups 7 and 8 from women who had progressive disease. The 40-45 kDa band was shown to be the P-chain of phosphatase haptoglobin by subjecting the electrophoretically-separated am specimens extracts to the Western-blotting procedure using an antin a Hilger haptoglobin antibody. Figure 4 shows representative results irers' recom-of this type of analysis for four patients who are not reice ranges for sponding to therapy. Specimens from three patients are the id 170 U ml-' same as those used in Figure 2 and the other group of t levels were specimens is from a patient used in Figure 3. It can be seen sFP EIA kits that the antibody detects a diffuse band at 40-45 kDa that trations were increases in intensity with increasing tumour involvement, Agency for and that these changes in antibody staining reflect those terial 72/225 observed for the 40-45 kDa bands observed on the silverace range was stained patterns. The lotus-extracted haptoglobin will subsethe x2 test.
quently be referred to as FHp. The sharp minor bands seen between 40 and 45 kDa on the silver-stained patterns, particularly in groups 1-3 Figure 2, did not blot for haptoglobin. quantitative estimations of the intensity of staining of the d six breast). FHp band as judged by two independent investigators. These ,pecimens are assessments are grouped according to the patient response at *e responding the time when the specimen was collected i.e., 'complete *es 2 and 3); response'; 'partial response or stable disease'; or 'progressive therapy and disease'. It can be seen that for both cancers, there is a highly -8, Figure 2 significant association between the amount of FHp present in t'here is an the blood and increasing tumour burden. In one ovarian tamring band patient (four specimens) and three breast patients (three te ovarian or specimens) FHp was undetectable even when they had progressive disease.
Twenty-six cancer patients (14 ovarian and 12 breast) were assessed for the presence of liver metastases using computed tomographic (CT) scans. Metastases could not be detected in 15 patients; whereas a moderate or strongly staining FHp was detectea in blooa specimens trom 13 of these patients.
Metastases were detected in the remaining 11, but only eight of these had a moderate or strongly staining FHp band. Statistical analysis of this data showed that there was no significant correlation between the presence of liver metastases and the amount of FHp in the blood (P >0.05, x2 test). In other studies, aspartate aminotransferase (AST) activity was measured in 91 specimens from 35 cancer patients (20 ovarian and 15 breast). Increased activity of this enzyme frequently indicates liver damage (Whitby et al., 1984). AST was elevated above the normal reference range (> 37 U m-') in only 11 of the specimens; whereas 54 specimens had a moderate or strongly-stained FHp band. Statistical analyses indicated that there was no correlation between the AST and FHp levels in the ovarian group (P >0.05, x2 test_ anti a vt-rv winlr neend-intinn hattivnasr fkaoa +--,-L%,atj-ajLjV a v 1i w a& uJbmaLLion LJeLween Lines two measurements in the breast group (P = 0.03, x2 test). Two other indicators of liver function (alkaline phosphatase and 6 bilirubin) were also measured in many of the blood speci-_ .. __ mens. Again, no significant correlation was observed between extracted sera from six women with breast cancer. were undergoing cytotoxic chemotherapy. Groups 1women who had stable disease and/or complete remis' 4-6 are from women who had progressive disease interval between the first and last samples for each 126, 220, 572, 112, 384 and 440 days respectively. s of lotus-All women Figure 4 A Western blot (anti-haptoglobin) of cancer sera lectin--3 are from extracts separated by electrophoresis. Groups 1 to 3 are the same sion; groups specimens as shown in groups 4 to 6 of Figure 2 respectively, e. The time except that only the first two specimens from group 5 are shown patient was in this figure. Group 4 are the first three specimens from group 5, Figure 3.
19 The association between FHp assessment and patient response to therapy is highly significant (ovarian, P <0.0001; breast, P = 0.002). The X2 test was carried out by pooling the data for the 'Moderate Band' and 'Strong Band' groups. the levels of these substances and the amount of FHp present.
FHp levels and alpha-fetoprotein concentrations were determined in 23 specimens from 16 patients with hepatocellular cancer. Five specimens had a weak band or no band, two of these being > 1,000 ig I-; 12 specimens had a moderate band, eight of these being > 1,000 g 1%'; and six specimens had a strong band, two of these being > 1,000 pg 1-'.
Statistical analysis of this data indicated that there was no significant correlation between the levels of these two substances (P >0.05, x2 test).

Discussion
This investigation has shown that there is an abnormallyfucosylated form of haptoglobin (FHp) in the blood of women with ovarian or breast cancer. This finding confirms the results from a previous preliminary study of 22 patients with different advanced cancers that included three women with breast cancer and three women with ovarian cancer (Thompson & Turner, 1987b). The 48 women evaluated in the present study were monitored for up to 22 months whilst receiving cytotoxic or hormone therapy. The intensity of the FHp band was found to increase with increasing disease involvement, and if a patient had no evidence of disease, i.e. 'a complete response', then either the FHp band was not detected or it was very weakly expressed. This pattern was consistent for 44 out of 48 of the patients studied. Four patients did not show an increase in FHp with progressive disease; a possible explanation for this will be discussed below. Although there was considerable variation in the therapeutic regime used from patient to patient, this variable did not seem to affect FHp levels.
As the liver is the major source of haptoglobin in the body (Koj, 1974) and as malignant tumours frequently metastasise to this organ (Weiss, 1985), it seemed possible that the FHp arose through the increasing presence of metastatic tumour in the liver that was affecting the fucosylation of haptoglobin. However, results from our study did not support this concept. There was a surprising lack of association between FHp expression and the presence of liver metastases on CT scans, and the elevation of serum enzymes that reflect liver damage. It is important to mention, however, that these latter methods only detect the presence of gross metastasis. They do not exclude the possibility that micrometastases were present, which were playing an important role in the production of FHp.
The effect of the tumour growing in the liver on the formation of FHp was further investigated by measuring the levels of FHp and AFP in the same specimens from individuals with primary hepatomas. AFP is a good marker for hepatomas, and blood levels are related approximately to tumour burden (Bates & Longo, 1987). There was no correlation between the levels of these two substances. In fact, in some patients who had high amounts of FHp the AFP level was low, and in other patients with high AFP the FHp levels were only moderately elevated. The most likely explanation is that as hepatomas grow and replace normal liver tissue, the ability of the liver to form FHp is reduced. This possibility is interesting because it suggests that only normal liver tissue can form abnormally fucosylated haptoglobin. It may explain why no strong relationship was detected between FHp levels and the presence of gross metastases as detected by the CT scans and the liver enzyme measurements. Also, three of the four patients who did not show an increase in FHp with progressive disease, had evidence of extensive liver metastasis at the start of specimen collection.
A possible source of FHp could be the tumour itself. Yoshimura et al. (1978) have shown that haptoglobin can be detected in the blood plasma of nude mice bearing a transplanted human renal cell carcinoma. Furthermore, a recent study (Kuhadja et al., 1989) has detected a protein in invasive breast carcinomas that is related to haptoglobin (Hpr). The HPR gene locus is a stretch of DNA located 2.2 kilobases downstream from the conventional HP locus (Maeda, 1985;Bensi et al., 1985). The HPR sequence codes for a protein whose a and P chains are distinct from, but highly homologous to haptoglobin 1-1. A major difference in the HPR sequence is the presence of a retrovirus-like element. One of the minor changes is the substitution of a histidine residue for a serine residue close to one of the glycosylation sites. As this change will switch-off the glycosylation at this site, Hpr will have only three potential sites of glycosylation compared with the normal four (Nilsson et al., 1981). Therefore, the lectin-binding properties of Hpr and Hp will probably be different, but whether FHp and Hpr are related is still unclear. In preliminary experiments, however, we have been unable to detect haptoglobin or FHp in cultured human cancer cell lines or in the media from these lines (unpublished observations).
A more likely source of FHp is the liver. The tumour may release soluble factors which promote the addition of fucose to haptoglobin in the liver. Recent evidence was suggested that cancer cells can produce factors (e.g. IL6) that can affect the glycosylation of acute phase proteins in the liver (Mackiewicz et al., 1989). As the factors could be secreted by tumour that was growing at the primary site, in the liver or in another secondary site, it might be expected that the production of FHp would be related to overall tumour burden. Thic conclusion agrees with our current findings.