We reported the expression of the homeodomain-containing transcription factor Engrailed-2 (EN2) in prostate cancer and showed that the presence of EN2 protein in the urine was highly predictive of prostate cancer. This study aimed to determine whether patients with prostate cancer have EN2 autoantibodies, what the prevalence of these antibodies is and whether they are associated with disease stage. The spontaneous immunoglobulin (Ig)G immune response against EN2 and for comparison the tumour antigen New York Esophageal Squamous Cell Carcinoma 1 (NY-ESO-1), were tested by enzyme-linked immunosorbent assay (ELISA) in three different cohorts of prostate cancer patients as well as a group of men genetically predisposed to prostate cancer. Thirty-two of 353 (9·1%) of the SUN cohort representing all stages of prostate cancer demonstrated EN2 IgG responses, 12 of 107 patients (11·2%) in the advanced prostate cancer patients showed responses, while only four of 121 patients (3·3%) with castrate-resistant prostate cancer showed EN2 autoantibodies. No significant responses were found in the predisposed group. Anti-EN2 IgG responses were significantly higher in patients with prostate cancer compared to healthy control males and similarly prevalent to anti-NY-ESO-1 responses. While EN2 autoantibodies are not a useful diagnostic or monitoring tool, EN2 immunogenicity provides the rationale to pursue studies using EN2 as an immunotherapeutic target.
Prostate cancer is the most common cancer in men in the United Kingdom, accounting for a quarter of all new cases of cancer in males . A large proportion of patients with localized small volume disease are either cured by treatment or do not require intervention (i.e. are managed by active surveillance). There are currently intense worldwide efforts to establish reliable biomarkers both for the early detection of prostate cancer and also to identify which cancers are likely to progress. We recently reported the diagnostic potential of Engrailed-2 (EN2) protein, a homeodomain-containing transcription factor secreted into the urine of men with prostate cancer . In a follow-up study we further demonstrated that urinary EN2 levels predict tumour volume in men undergoing radical prostatectomy for prostate cancer  and thus may be useful in deciding which patients to treat and which to monitor closely. Moreover, tissue expression (cytoplasmic, nuclear and membrane) of EN2 was found exclusively in prostate cancer, and not normal prostatic epithelium in tissue microarrays and patient specimens. EN2 was expressed equally at all stages of prostate cancer (T stages 1–4) as well as in metastatic disease (lymph nodes) (Morgan, personal communication).
Antibodies fulfil their adaptive immune response commonly to prevent infection. However, the function of tumour-associated autoantibodies, which bind self-antigens and have escaped self-tolerance, is generally unknown. Autoantibodies can mediate antibody-dependent cell-mediated cytotoxicity and complement-dependent cytotoxicity. They may also enhance antigen cross-presentation and activation of T lymphocytes. It has been shown that immunoglobulin (Ig)G responses to tumour-associated proteins exist in cancer patients, making it possible to use human sera to identify immunologically recognized potential tumour antigens . IgG responses to tissue-associated antigens may have diagnostic value, aid in risk stratification in terms of tumour progression or provide a tool for monitoring patients response to treatment. Previous studies have clearly shown a multitude of antibody responses to prostate-associated antigens [5-8]. We proposed that de-novo expression of EN2 in prostate cancer could potentially elicit an autoantibody response, and that this could be quantified and potentially used as an adjunct to current diagnostic and monitoring biomarkers.
In this study we assessed the spontaneous antibody response to EN2 in men with prostate cancer, both treatment-naive at different stages but also in patients who had received optimal treatment. We detected a significantly higher prevalence of IgG responses to EN2 in prostate cancer patients compared to healthy control males, with positive antibody responses being found predominantly in patients with higher volume and more advanced stage prostate cancer.
Materials and methods
The study evaluated blood samples from cancer patients, individuals at high risk of developing prostate cancer and healthy volunteer controls. In all cases, written informed consent was obtained and the donation of blood samples for autoantibody evaluation had received approval by the local ethics committee.
Our control groups comprised men who were unlikely to harbour prostate cancer based on the absence of symptoms, no family history of cancer and a prostate-specific antigen (PSA) <2 ng/ml (average age 65 years; range 40–89). In a similar manner, the female controls for the breast and ovarian cancer study were women with no past history of any cancer, no symptoms and no family history of any malignancy. Patients with histological (on routine biopsy) evidence of benign prostatic hypertrophy, prostatic intra-epithelial neoplasia and prostatitis were excluded from the study. These patients generally have an abnormal PSA, and even with saturation biopsies an underlying malignancy cannot be excluded. In the case of prostatitis the PSA is often markedly raised, and may remain so for 6–9 months. Despite our large cohorts of patients and controls in the study, the total number of patients with these three conditions would have been too small for meaningful statistical comparison.
Blood was taken from prostate cancer patients from the Royal Surrey County Hospital (Guildford, UK). We evaluated four specific cohorts, as follows.
Samples from a comprehensive prostate cancer biobank, consisting of the pretreatment sample of 353 men with all stages of prostate cancer (SUN cohort), LREC: 08/H1306/115. This cohort of patients consisted of men with newly diagnosed cancer patients prior to starting any form of treatment. The median age of the cohort was 71 years, the median PSA at diagnosis was 9 and the median Gleason score was 6.
Patients with rising PSA despite luteinizing hormone-releasing hormone (LHRH) agonist hormone treatment with normal radiological evaluation and no symptoms, designated CRRPSA (castrate-resistant, rising PSA). This was a cohort of patients recruited to a clinical trial of allogeneic whole cell vaccine, EUDRACT number 2006-001619-31. The median age of the patients was 71 years, the median PSA was 5·5 and the median Gleason score was 7.
Men with castrate-resistant disease with radiological evidence of metastatic disease, LREC: 08/H1306/115. This was a cohort of patients presenting to oncology clinics and receiving chemotherapy. All patients had metastatic prostate cancer with either bone or soft tissue disease.
Patients at high risk of cancer. Samples from the IMPACT cohort (Identification of Men with a genetic predisposition to Prostate Cancer: Targeted Screening in BRCA1/2 mutation carriers) were used. To be eligible for this cohort, men had to be either a breast cancer 1 (BRCA1) or BRCA2 mutation carrier or from a family harbouring a BRCA1/2 mutation but had tested negative themselves (LREC: 05/MRE07/25) .
For control comparison of unrelated cancers, we tested two independent female breast cancer sample sets of sera taken from women prior to first definitive treatment for their breast cancer from Charing Cross Hospital (LREC: 07/MRE09/54) and Southampton General Hospital (LREC: 05/Q1702/13) . We also evaluated a further control group of women with ovarian cancer at the Royal Surrey County Hospital (LREC: 09/H1103/50). Most women had advanced ovarian cancer (FIGO stages 3c or 4) and had received three cycles of neoadjuvant chemotherapy prior to their blood test. In a few instances, blood was obtained prior to any treatment. Blood samples from each cohort were taken in accordance with the protocol associated with each study. In all patients, sera/plasma was stored in 0·5 ml aliquots at −80°C until used for analysis.
Expression of recombinant EN2 and New York esophageal squamous cell carcinoma 1 (NY-ESO-1) proteins within non-fusion and fusion Escherichia coli expression vectors
EN2 cDNA with optimized codon usage was synthesized and cloned by GenScript, USA. This EN2 cDNA was recloned into the pQE 31 plasmid (Qiagen, Manchester, UK), placing it under the control of a T7 promoter as well as adding an aminoterminal histidine tag. The NY-ESO-1 cDNA was obtained from pLPCX NY-ESO-1 (a gift from Beatrice Yin, Sloan Kettering Institute, NY, USA). To obtain high expression of NY-ESO-1 protein a fusion partner (mouse dihydrofolate reductase) was needed. To achieve this expression, the cDNA of NY-ESO-1 was recloned into the pQE 40 plasmid (Qiagen), which resulted in the fusion of NY-ESO-1 to the carboxyterminal end of mouse dihydrofolate reductase (plus histidine tag). As a non-specific E. coli protein control, mouse dihydrofolate reductase was produced from the pQE 40 plasmid containing no insert. All plasmids were transformed into E. coli host strain (M15; Qiagen), which contains multiple copies of the pREP4 plasmid which constitutively expresses the lac repressor protein, resulting in regulated efficient and controlled expression of E. coli proteins. The histidine tags allow purification of recombinant proteins by affinity chromatography on Ni-NTA resin (Qiagen) under denatured conditions.
Enzyme-linked immunosorbent assay (ELISA) to detect circulating autoantibodies to EN2
ELISA MaxiSorp 96-well plates (Thermo Scientific/Nunc, Loughborough, UK) were coated with recombinant EN2, NYESO-1 or control DHFR protein produced in E. coli at a concentration of 4 ug/well in 0·1 M carbonate buffer (33·5 mM Na2CO3 0·1 M NaHCO3, pH 9·6) and incubated overnight at 4°C. Plates were blocked with 3% bovine serum albumin/phosphate-buffered saline (BSA/PBS) for 2 h at room temperature on a rapid shaker. All washes were performed with PBS/0·1% Tween-20. Following four washes, plates were incubated with patients' and control plasma diluted 1:100 in 3% BSA/PBS for 1 h at room temperature on a rapid shaker. All plasma/sera were assayed in triplicate. Plates were washed and incubated with goat anti-human IgG conjugated to horseradish peroxidase (Jackson ImmunoResearch, Newmarket, UK) or mouse anti-human IgG1, IgG2, IgG3 or IgG4-horseradish peroxidase (HRP) (Life Technologies, Paisley, UK) in 3% BSA/PBS for 1 h on a shaker at room temperature. The plates were then incubated with 100 μl of 3,3′,5,5′-tetramethylbenzidine solution (TMB) (Sigma, Poole, UK) for colour development. The reaction was stopped by addition of 2 M H2SO4. Optical density (OD) was measured at 450 nm by using a plate reader (Beckman Coulter, High Wycombe, UK). The mean value plus 3 standard deviations calculated from the test results for healthy controls for plasma IgG was used as a cut-off value for evaluation of each assay. This ELISA was standardized using rabbit (EN2; Abcam, Cambridge, UK) and goat antibodies (NY-ESO-1; Lifespan BioSciences, Seattle, WA, USA) with specificity for each of the antigens.
To confirm ELISA results, EN2 detection was performed by running 1 μg of purified EN2 protein or the control protein dihydrofolate reductase (DHFR) on a NuPAGE gel followed by semi-dry Western blotting, blocking using 5% Marvel and chemiluminescent detection. EN2 detection was achieved using 1:1000 dilution HRP-conjugated donkey anti-goat IgG (Jackson ImmunoResearch). To show that plasma or sera antibodies also recognize our recombinant EN2 protein, blots were incubated with patient sera or plasma diluted 1:100 for 24 h, washed and detected using HRP-conjugated goat anti-human IgG (Jackson ImmunoResearch) diluted 1:10 000.
The χ2 test with a 95% confidence interval was performed using MedCalc to evaluate the statistical significance of differences in the proportion of EN2 or NY-ESO-1 antibody responders between patient cohorts and healthy controls. Differences were considered significant when the probability values obtained from the statistical tests were 0·05 or less.
In order to determine whether circulating EN2-specific antibodies were present in patients and healthy controls, an ELISA was developed using recombinant EN2 (rEN2) protein produced by expression in E. coli. Samples from patients and healthy controls were screened on rEN2 and as a comparison on one of the most immunogenic cancer testis antigens, recombinant NY-ESO-1 (rNY-ESO-1)-coated microplates. In addition, samples were also screened against a non-immunogenic murine protein (DHFR) similarly produced in E. coli to eliminate sera/plasma that reacted non-specifically with bacterial contaminants in the preparation. Western blotting experiments were performed to verify ELISA results in selected patient samples compared to a commercial antibody. The commercial antibody detected a ∼45 kDa protein only in the lane containing the purified recombinant EN2 protein (Fig. 1, goat anti-EN2 antibody; Abcam). Similarly, patient sera and plasma samples that were shown to be EN2 antibody-positive by ELISA also gave a positive signal by Western blotting (Fig. 1, patients 1, 2 and 3) and conversely, those patients who were negative for EN2 antibodies by ELISA gave a negative result by Western blotting (Fig. 1, patients 4 and 5).
Patients with prostate cancer demonstrate spontaneous IgG anti-EN2 responses in plasma/serum
We first evaluated IgG anti-EN2 and anti-NY-ESO-1 autoantibody responses in our SUN cohort. These were pretreatment samples from men with all stages of prostate cancer (Fig. 2a). To be considered positive for an IgG response the cut-off values were calculated as the mean value plus three standard deviations from the data pool of the control sera/plasma samples. We found that 32 of 353 (9·1%) of the entire SUN cohort taken together as one group demonstrated EN2 IgG responses, compared to a 4·8% response to NY-ESO-1. The proportion of EN2 IgG responders was significantly different between patients and controls (P < 0·001). Figure 2b shows titration curves of EN2 antibody-positive and -negative sera from the SUN cohort. Similar titration curves were also obtained from NY-ESO-1 antibody-positive and -negative sera (data not shown). If patients were subdivided according to Gleason score or clinical stage (Fig. 2c,d), there were no significant differences in the proportion of EN2 IgG responders with respect to disease stage (T1–4 individually) or Gleason score. T4 stage patients were not considered statistically because of the low numbers. In addition, there was a significant difference in the proportion of NY-ESO-1 IgG responders from the whole SUN cohort compared to healthy controls (P < 0·05). However, similar to the EN2 IgG response, there were no significant differences in the proportion of NY-ESO-1 IgG responders with respect to disease stage (T1–4 individually) or Gleason score.
We next evaluated samples from men with a rising PSA level despite optimal hormone treatment, but with no disease radiologically (CRRPSA). We found no significant increase in the proportion of EN2 IgG responders in this group of patients compared to healthy controls. However, the proportion of patients within this group responding to NY-ESO-1 was statistically different from healthy controls (P < 0·01) (Fig. 3a).
We found that in men with castrate-resistant advanced metastatic disease 12 of 107 patients (11·2%) were confirmed as positive for IgG responses to EN2. This proportion of EN2 IgG responders was statistically different compared to healthy controls (P < 0·001). None of these patients were receiving chemotherapy at the time the blood was drawn (Fig. 3b).
In men at high risk of developing prostate cancer (IMPACT study participants), no significantly increased IgG responses to EN2 or NY-ESO-1 were detected compared to the age-matched healthy control group (Fig. 3c).
Of the patients who demonstrated a positive IgG response to EN2 and/or NY-ESO-1, only two of the 47 patients from the SUN study cohort showed a response to both antigens, none of the antibody-positive patients from the CRRPSA cohort responded to both antigens and only two of 19 patients showed a response to both antigens from the men with castrate-resistant disease with radiological evidence of metastatic disease.
Dominant IgG1 and IgG3 response against EN2
Twenty-one serum/plasma samples reacting with EN2 were additionally subjected to IgG subclass testing (IgG subclasses 1–4). The level of the IgG subclass of the EN2 antibody was determined in each patient using the same ELISA as before, but for detection we used subclass-specific secondary antibodies. We predominantly found antibodies from subclass IgG1 and/or IgG3 in all patients. Eleven of 21 patients tested showed both an IgG1 and IgG3 response, nine of 21 patients gave only an IgG1 response and one patient showed only an IgG3 response. No significant IgG2 or IgG4 response was observed in any of the patients (Fig. 4).
EN2 antibody response remains stable in seropositive patients regardless of treatment
From the SUN study cohort sequential follow-up, samples were available which allowed us to study the EN2 autoantibody response in a longitudinal manner. From those patients who had demonstrated a positive autoantibody response to EN2, sequential samples were available from five patients under active surveillance, three patients who had received radiotherapy and five patients who had received a radical prostatectomy. ELISA results on the sequential samples from these patients showed that the antibody response remained stable in all but one of these patients regardless of treatment and over a period of up to 3 years 11 months. In only one of the patients (SUN 147) who had received a radical prostatectomy did the EN2 antibody response fall below the cut-off level for positivity following surgery and remained negative for 2 years 10 months post-surgery (Fig. 5).
No significant detection of anti-EN2 IgG responses in female oncological patients
The assessment of serum/plasma IgG anti-EN2 antibodies in patients with malignancies was extended to female patients with breast or ovarian cancer, cancer types which we have found to have a high prevalence of EN2 expression at the protein level (data not shown). As shown in Fig. 6, autoantibody was rarely detected and was not higher in patients with breast or ovarian cancer compared to female healthy controls. This was in contrast to IgG anti-NY-ESO-1 responses, which were found clearly in both female cancer groups at a frequency published previously in the literature .
Serum autoantibodies directed against tumour-associated antigens have been detected frequently in the sera of patients with different types of cancer , including prostate cancer. A multitude of autoantibodies to various antigens such as PSA, prostatic acid phosphatase, human epidermal growth factor receptor 2 (HER-2)/neu, p53, glucose-regulated protein 78 and alpha-methylacyl-CoA racemase (AMACR) have been observed in the sera of prostate cancer patients [12-15]. Several groups have tried to identify antibody responses to such prostate tissue antigens as a means of prostate cancer detection. Maricque et al. showed that IgG responses to a panel of commonly recognized prostate antigens were detected as frequently in patients with clinical prostatitis as in patients with prostate cancer, thus demonstrating the need to define those antigens which are clinically relevant for prostate cancer . Other studies investigating the antibody responses to individual proteins, including huntingtin-interacting protein 1 (HIP1)  and AMACR , have shown that antibody responses to these proteins are highly specific for prostate cancer. Thus the identification of new immunogenic antigens specific for prostate cancer will help to build up a panel of commonly recognized prostate cancer antigens that could be used either to detect prostate cancer or to evaluate immune responses to the prostate. Indeed, Smith et al. used a panel of 126 prostate tissue-associated antigens common in patients with prostate cancer to investigate whether changes in IgG responses to this panel might be used as a measure of immune response, and potentially antigen spread, following prostate cancer-directed immune-active therapies . They found that different antigens were recognized following androgen deprivation compared with vaccine therapies.
In the current study we assessed the levels of autoantibodies against EN2 in sera from patients with prostate cancer, other cancers and from individuals in high-risk groups. This is, to our knowledge, the first assessment of spontaneous immune responses to EN2 in prostate cancer. We have demonstrated a significant increase in anti-EN2 IgG antibodies in patients with prostate cancer compared to control subjects. Furthermore, the frequency of anti-EN2 IgG responses in our prostate cancer patient cohorts was similar to the number of seropositive patients specific for the highly immunogenic tumour antigen, NY-ESO-1. NY-ESO-1 is a cancer/testis antigen which is expressed in a number of human cancers, including prostate cancer . NY-ESO-1 appears to be one of the most immunogenic tumour antigens inducing a humoral immune response and specific CD8 T cell reactivity in ∼50% of patients with advanced NY-ESO-1-expressing tumours [18, 19]. Due to its high immunogenicity and broad expression in a variety of cancers, monitoring of antibody responses to NY-ESO-1 has already been used as a selection criterion for patients enrolling in NY-ESO-1 cancer vaccine trials and as a tool to assess the immunological and clinical responses of the treatment [18-20]. Notably, an elevated EN2 IgG response was not related to an elevated NY-ESO-1 response, indicating that a high EN2 response is highly specific and not related to a general autoimmune phenomenon. The prevalence of humoral immune responses to EN2 within our prostate cancer patient cohorts was similar to that reported previously in prostate cancer patients for other prostate cancer-associated proteins (11% PSA and 15·5% Her-2/neu) .
We first evaluated pretreatment sera from men representative of stages 1–4 prostate cancer from a prospective biobank. We found a significant prevalence of IgG responses to EN2 in the cancer patients versus controls, and this was comparable to the response frequency to NY-ESO-1 measured in the same samples. As this group contained patients at different stages of prostate cancer, we then evaluated whether IgG responses to EN2 were more frequent at a particular disease stage. We observed no statistical difference in the frequency of EN2 IgG responses between groups of patients based either on Gleason score or T stage. This is in keeping with studies by Maricque et al., who also reported that immune responses to multiple antigens were detectable as frequently in patients with early-stage disease as in patients with advanced metastatic disease .
We then went on to test two groups of patients with castrate-resistant disease. Here we found that there was a significantly lower proportion of EN2 IgG responders in those patients with undetectable cancer but a rising PSA (CRRPSA). In contrast, the group with metastatic disease displayed EN2 IgG responses comparable to the SUN cohort. This suggests that detectable disease needs to be present in order for a measurable EN2 IgG response to be present. This is supported further by our finding that there were no significantly elevated anti-EN2 IgG responses or, indeed, anti-NY-ESO-1 IgG responses in the IMPACT high-risk group compared to healthy controls. A number of studies have reported that circulating autoantibodies to tumour-associated antigens have been found to be detectable several years before radiographic detection or incidence screening was able to identify the tumours [21-23]. However, the mechanism underlying their emergence and regulation of their production is not fully understood. Previous reports have described the detection of specific serum autoantibodies in prediagnostic sera and hence their potential use as a screening tool in asymptomatic high-risk populations [21-23]. Due to the availability of sequential samples from patients within the SUN study cohort, we were able to address this by performing a longitudinal assessment of the EN2 autoantibody response in seropositive patients and the effects of different types of treatment on the antibody response. Interestingly, in all patients the EN2 autoantibody response persisted over the entire sampling period (up to nearly 4 years in some patients) regardless of the type of treatment received. The only exception was one patient who became EN2 seronegative following radical prostatectomy. This finding is perhaps not surprising, as it is known that low-level antibodies can persist up to a lifetime in an individual following vaccination or infection. This can be attributed to ongoing production from a rapid turnover of memory B cell pools  or long-lived plasma cells  which can be maintained by antigenic stimulation (e.g. persistent antigenic re-exposure) and/or antigen-independent activation mechanisms . One can speculate whether the persistence of the antibody response is due to low levels of residual tumour or simply to long-lived plasma cells. It is tempting to conclude that the single patient in whom the antibody response disappeared following surgery was actually cured, given the stability of the antibody levels in the majority of patients studied. However, the persistence of the EN2 antibody response in most patients precludes its use as a monitoring tool for response to treatment.
In this study, we went on to further characterize the humoral response to EN2 by determining the subclass of antibody induced. It is known that human B cells produce four subclasses of IgG (IgG1, IgG2, IgG3, IgG4), with each subclass having different biological functions [27, 28]. The dominant subclasses induced in the antibody response against EN2 were IgG1 and IgG3. Although the clinical significance of the different IgG1 and IgG3 patterns of response observed in our patients cannot be addressed due to the small numbers of patients studied, it is known that the distribution of IgG subclasses within an antibody response is affected by a number of factors, including nature of the antigen, dose, route of entry and also the host genotype . This was demonstrated in a vaccination study using recombinant carcinoembryonic antigen (CEA), where a gradual shift was seen from a predominant IgG1 response at 6 months to an IgG4 response at 15 months . Similar to our findings, the humoral response to NY-ESO-1 has also been reported to consist of antibodies of the IgG1 and IgG3 subclass . IgG1 induction results from T helper type 1 (Th1) CD4 T cell activation , while immune-complexes consisting of the antigen and IgG1 or IgG3 bind mainly to FcγRIIa [33, 34], which is a potent leucocyte activator and stimulates the release of high levels of inflammatory cytokines. Thus the induction of IgG1 and IgG3 provide favourable conditions for effector T cell activation in the immune response to EN2. Our group are currently investigating cellular immune responses to EN2.
Interestingly, the present work showed that the increased anti-EN2 IgG levels appeared to be more significant in male cancer patients than in female cancer patients. This was in contrast to the autoantibody response to NY-ESO-1, which was significant in both male and female cancer patients. This gender difference in the humoral immune response has been reported before, and further supports the suggestion that sex hormones may significantly affect the humoral immune response to some tumour-associated antigens [35, 36].
In summary, this study has shown that the homeodomain-containing transcription factor EN2 is immunogenic, as evidenced by the significant prevalence of IgG responses to EN2 in prostate cancer patients. Antibody immunity to this tumour antigen was observed at different stages of the disease. The confirmation of its immunogenicity provides the rationale to pursue studies using EN2 as an immunotherapeutic target.
Furthermore, our findings may have additional clinical relevance, as it is unlikely that any single marker for prostate cancer will have the desired high specificity and sensitivity for initial diagnosis. Combined analyses of different autoantibodies have already increased diagnostic sensitivity and specificity in breast cancer [p53, HER2/neu, insulin-like growth factor binding protein 2 (IGFBP-2) and topoisomerase 2-alpha (TOPO2α)]  and lung cancer (p53, NY-ESO, CAGE, GBU4-5, annexin 1) . Identification of antibodies to further novel tumour-associated proteins may thus help to build up a ‘panel’ which may also include non-protein biomarkers and which, in combination, may be sufficiently predictive.
We would like to thank The Prostate Project Charity for their financial support.
N. E. designed the study, carried out experiments and wrote the manuscript; G. S. produced recombinant proteins and performed Western blotting; M. D. carried out ELISAs; S. M. carried out ELISAs; G. F. carried out ELISAs; E. K. helped in acquiring patient samples; R. E. helped in acquiring patient samples; J. S. helped in acquiring patient samples; D. P. helped in acquiring patient samples; R. C. helped in acquiring patient samples; N. M. helped in acquiring patient samples; A. M. helped in acquiring patient samples; H. P. was involved in the study design.