Clinical Chemistry Siemens Point of Care - Urinalysis
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Clinical Chemistry 54: 713-722, 2008. First published February 15, 2008; 10.1373/clinchem.2007.096792
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
clinchem.2007.096792v1
54/4/713    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Petrik, V.
Right arrow Articles by Papadopoulos, M. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Petrik, V.
Right arrow Articles by Papadopoulos, M. C.
Related Collections
Right arrow Proteomics and Protein Markers
(Clinical Chemistry. 2008;54:713-722.)
© 2008 American Association for Clinical Chemistry, Inc.

Proteomics and Protein Markers

Serum {alpha}2-HS Glycoprotein Predicts Survival in Patients with Glioblastoma

Vladimir Petrik1, Samira Saadoun1, Alison Loosemore1, Josie Hobbs2, Kirstie S. Opstad4, Joanna Sheldon2, Edward Tarelli3, Franklyn A. Howe4, B. Anthony Bell1 and Marios C. Papadopoulos1,a

1 Academic Neurosurgery Unit, 2 Protein Reference Unit, 3 Medical Biomics Centre, and 4 Biomedical Magnetic Resonance Research Group, St. George’s University of London, Cranmer Terrace, London, U.K.

aAddress correspondence to this author at: Academic Neurosurgery Unit, Jenner Wing Room 1.122, St. George’s University of London, Cranmer Terrace, London, SW17 0RE, UK. Fax +44–20-87255139; e-mail mpapadop{at}sgul.ac.uk.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Background: Glioblastoma, the most common primary brain tumor, has variable prognosis. We aimed to identify serum biomarkers that predict survival of patients with glioblastoma.

Methods: In phase 1 (biomarker discovery), SELDI-TOF mass spectra were studied in 200 serum samples from 58 control subjects and 36 patients with grade II astrocytoma, 15 with anaplastic astrocytoma, and 91 with glioblastoma. To identify potential biomarkers, we searched for peptide peaks that changed progressively in size with increasing malignancy. One peak, identified as the B-chain of {alpha}2-Heremans-Schmid glycoprotein (AHSG), was less prominent with increasing tumor grade. We therefore investigated AHSG as a survival predictor in glioblastoma. We measured serum AHSG by turbidimetry and determined indices of malignancy, including tumor proliferation (Ki67 immunolabel) and necrosis (tumor lipids on magnetic resonance spectroscopy). In phase 2 (biomarker validation), the prognostic power of AHSG was validated in an independent group of 72 glioblastoma patients.

Results: Median survival was longer (51 vs 29 weeks) in glioblastoma patients with normal vs low serum AHSG concentrations (hazard ratio 2.7, 95% CI 1.5–5.0, P <0.001), independent of age and Karnofsky score. Serum AHSG inversely correlated with Ki-67 immunolabeling and tumor lipids. A prognostic index combining serum AHSG with patient age and Karnofsky score separated glioblastoma patients with short (<3 months) and long (>2 years) median survival. The prognostic value of serum AHSG was validated in a different cohort of glioblastoma patients.

Conclusions: We conclude that serum AHSG concentration, measured before starting treatment, predicts survival in patients with glioblastoma.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Glioblastoma multiforme (GBM,1 grade IV astrocytoma) is the most common and most aggressive brain tumor in adults(1). The prognosis of individual patients is variable; about 50% of GBM patients die within 30–40 weeks of diagnosis, but 10% live more than 2 years(2)(3). Survival predictors available before surgery are clinically useful for deciding whether to offer radical excision and adjuvant treatment vs palliative care and for counseling patients and their families. At present, the only established preoperative survival predictors in patients with GBM are age and Karnofsky performance score(2)(3)(4).

The aim of this study was to identify serum biomarkers that improve survival prediction in patients with GBM. We searched the serum proteome of astrocytoma patients and control subjects using SELDI-TOF mass spectrometry (MS), which has successfully identified serum biomarkers in other forms of cancer(5)(6) and infection(7). A major advantage of SELDI-TOF MS is that it makes no a priori assumptions about the nature of biomarkers.

We hypothesized that any peptide peaks that progressively change with increasing malignancy (control -> grade II astrocytoma -> anaplastic astrocytoma -> GBM) are potential survival biomarkers for patients with GBM. The only peak that satisfied this condition corresponded to the B-chain of {alpha}2-Heremans-Schmid glycoprotein (AHSG, fetuin-A). AHSG, 1 of the most abundant plasma proteins, consists of 2 polypeptide chains (A and B), linked by disulfides(8). Using survival analysis, Ki67 immunohistochemistry, and magnetic resonance spectroscopy (MRS), we characterized several properties of AHSG in patients with GBM. We propose a novel prognostic score that illustrates the clinical potential of serum AHSG levels in GBM patients.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
patients
The project was approved by the Wandsworth and King’s College Hospital Research Ethics Committees. For part 1 (biomarker discovery), 228 subjects were recruited from St. George’s Hospital between 2002 and 2004 after informed consent. For part 2 (biomarker validation), 72 additional patients with GBM were recruited (42 from St. George’s Hospital and 30 from King’s College Hospital) in 2006. These neurosurgical units cover different regions in London of about 2.5 million (St. George’s) and 3.0 million (King’s) people. One blood sample per tumor patient was obtained on the day before surgery. Survival time was measured from the date of sample collection. In 7 patients, samples were taken before and after starting dexamethasone.

sample size estimates
The study has power 0.8 and {alpha} = 0.05 given 2 years of patient follow-up. We estimated a sample size of 88 GBM patients by assuming that the putative prognostic biomarker separates patients into 2 groups, 75% with a median survival of 6 months vs 25% with median survival of 1 year. Ninety-one GBM patients were recruited; 10 of the 91 were followed for >2 years. We also included 109 other patients (controls and grade II and III astrocytomas) to determine if the serum level of the putative biomarker progressively changes with increasing malignancy.

patient recruitment
Study inclusion criteria for the astrocytoma patients included neurosurgical unit admission, surgical treatment plan, ability to consent, age 18–80 years, and no previous central nervous system (CNS) disease. Control subjects had no CNS pathology and no known tumor or systemic disease; they are patients with lumbar radiculopathy and hospital staff. One hundred forty-two astrocytoma patients and 58 control subjects admitted to St. George’s Hospital were recruited for phase 1 (biomarker discovery). For phase 2 (biomarker validation), we recruited 72 patients with GBM from St. George’s and King’s College hospitals. Twenty-eight patients with nonglioma brain diseases or systemic tumors were used to determine whether the biomarker is GBM specific. All subjects approached agreed to donate samples, and all recruited subjects were included in the study. Every tumor patient had surgery and histological diagnosis. We followed up patients with GBM for at least 2 years (phase 1) or 30 weeks (phase 2).

clinical scales
Karnofsky performance(9) and Charlson comorbidity(10) indices were determined by consensus between a specialist nurse and a neurosurgical resident. A Karnofsky score ≥70 signifies ability to self-care and a high Charlson score indicates significant comorbidity. An astrocytoma with no comorbidity has a Charlson score of 2.

treatments
All patients with GBM and anaplastic astrocytoma had surgery, followed by radiotherapy and chemotherapy. Surgery was craniotomy and debulking of >80% of the tumor mass. Radiotherapy (60 Gy) was given 5–6 weeks postoperatively, and PCV chemotherapy (procarbazine/lomustine/vincristine) was administered when the tumor recurred after radiotherapy. None of the patients had resection of recurrent tumor or Gliadel implants.

mass spectrometry
Serum was processed on CM10 chips as described(7). The chips were placed in a PBS-II mass spectrometer (Ciphergen Biosystems) and TOF mass spectra (0–100 kDa) were generated at laser intensities 220, 240, and 260. We minimized the deviation of the total ion current (0–100 kDa) to within 0.4–2.5 x the mean of all patients(7). Biomarker Wizard version 3.1 (Ciphergen Biosystems) identified corresponding peaks in each spectrum within 0.3% of the molecular weight. Coefficients of variation were 10.5% (intraassay) and 20.6% (interassay) for peak height and 0.05% (intraassay and interassay) for mass-to-charge ratios.

2.740 Kda peptide identification
We sequenced the 2.740-kDa peptide using tandem MS and used immunodepletion to prove that the 2.740-kDa peak was derived from AHSG. Rabbit polyclonal anti-AHSG IgG antibody (10 µg; Biovendor) was linked to protein G Sepharose beads (20 µL 50% suspension in PBS, 25 °C, 1 h; Sigma-Aldrich). Diluted nonreduced serum (10 µL, 1:20 in PBS) was incubated with 30 µL antibody-bead complex suspension (90 min, 25 °C). After centrifugation at 3000g, the supernatant (AHSG-depleted serum) and pellet (AHSG-antibody-bead complex) were analyzed by SELDI-TOF MS.

ahsg and hscrp assays
We measured serum AHSG by use of an automated turbidimetric method on the Roche Cobas Mira using polyclonal rabbit anti-AHSG (Dade Behring) and calibrated against specific protein calibrants (SPS-01). Coefficients of variation were <5% at low, medium, and high concentrations. We measured high-sensitivity C-reactive protein (hsCRP) with an in-house sandwich ELISA using goat polyclonal antisera to human CRP (Diasorin) and horseradish peroxidase (HRP)-conjugated polyclonal rabbit antihuman CRP antisera (Dako). The method was calibrated against the international reference preparation CRM470. Coefficients of variation were <7% at normal and raised concentrations.

immunohistochemistry
Tissue sections were incubated with rabbit anti-Ki67 (1:1000, 25 °C, 1 h; Novocastra) followed by biotinylated secondary antibody (1:1000, 1 h), and avidin-HRP (Vector). Immunolabeling was brown (diaminobenzidine [DAB]). Omitting the primary antibody produced no staining. Two observers unaware of the histological diagnosis quantified the immunostaining by consensus. The percentage of immunostained nuclei was calculated in 5 high-power fields. Necrotic areas and blood vessel/leukocyte staining were excluded.

mrs
1H MR spectra were acquired and analyzed as described(11) on a 1.5-T Signa MR system (GE) using the automated Probe-P acquisition method with a 30-ms echo time. Spectra were analyzed with LC Model version 5.2-3S to obtain metabolite and lipid concentrations using the tumor water signal as a reference. The biochemicals quantified were: total choline, total creatine, myoinositol, N-acetyl-aspartate, and lipids with peaks at 1.3 and 0.9 ppm. Cellular lipids (marker of necrosis) rise, whereas myoinositol (predominantly from glia) and N-acetyl-aspartate (neuronal marker) fall, with increasing grade of astrocytoma(12).

statistical analysis
We tested peptide peaks for progressive change with increasing tumor grade using Spearman’s correlation. We performed 2-group comparisons with Student t or Mann-Whitney U tests. We carried out survival analysis using log-rank tests and used Cox regression proportional hazard modeling with forward stepwise analysis to determine the contribution of several variables to survival. We investigated the relation between serum AHSG and Ki67 labeling or MRS peaks using Spearman or Pearson correlation analysis. Data are expressed as mean (SE) or 95% CI, and all P values are 2-sided. Calculations were done with XLStat version 7.1 (Addinsoft) and SPSS version 12.0 (SPSS).


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
patients
Patient details are summarized in Table 1 . With increasing tumor grade, we noted a progressive rise in mean patient age, a fall in mean Karnofsky score, a rise in mean comorbidity score (Charlson), and reduced median survival. Malignant astrocytomas were more common in men. These characteristics are comparable with other clinical studies for age and sex distributions of patients with grade II(13) and III and IV(14) astrocytomas and survival of grade III and IV astrocytoma patients(2).


View this table:
[in this window]
[in a new window]

  		Table 1. Baseline characteristics of control and astrocytoma patients.

analysis of spectra
Normalized and calibrated spectra were analyzed with Ciphergen’s Biomarker Wizard to produce 192 peak clusters. Our selection criterion was that potential biomarkers should progressively differ in size (P <0.01) between control and grade II, III, and IV astrocytoma groups. One peak (2.740 kDa) was found in progressively diminishing concentration in the sera of control [50.2 (3.3)], grade II [38.5 (3.1)], grade III [27.0 (4.6)], and grade IV [24.5 (1.7)] astrocytoma (Fig. 1A ). The 2.740-kDa peak was therefore investigated as a potential biomarker of survival in patients with GBM.


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Figure 1. Serum concentration of 2.740-kDa peptide.

(A), Representative SELDI-TOF spectra (2–10 kDa) of 2 control patients (Con), 2 patients with grade II astrocytoma, and 2 patients with grade IV astrocytoma. Arrowheads mark the 2.740-kDa peak. (B), Size of the 2.740-kDa peak in 58 controls and 36 grade II, 15 grade III, and 91 grade IV astrocytoma patients. Mean ± SE. Significant correlation between 2.740-kDa peak size and tumor grade (r = –0.50, 95% CI –0.60 to –0.38, P <0.0001).

2.740-Kda peak and survival
No control subject had a 2.740-kDa peak size <22, whereas 15% of grade II, 40% of grade III, and 48% of grade IV astrocytoma patients did (Fig. 1BUp ). We used a peak size of 22 as an arbitrary cutoff to divide the 91 patients with GBM into 2 groups: 44 patients had 2.740-kDa peaks <22 and a median survival of 27 weeks vs 37 weeks for 47 patients with 2.740-kDa peaks ≥22 (hazard ratio [HR] 1.5, 95% CI 1.1–2.5, P <0.05).

properties of 2.740-Kda peak
The intensity of the 2.740-kDa peak did not correlate with patient age (r = 0.03, P = 0.81, n = 200), sex [37.1 (2.3) for 112 men; 32.8 (2.0) for 88 women; P = 0.16], Karnofsky score (r = 0.08, P = 0.43, n = 200), or Charlson score (r = –0.04, P = 0.66, n = 200). The 2.740-kDa peptide concentration was not reduced by steroids in 7 people with serum samples taken before and ≥72 h after starting steroid treatment [33.2 (4.4) presteroid vs 37.2 (5.2) poststeroid, P = 0.63], consistent with a recent report(15). Peak size [46.0 (2.6), n = 33] was unchanged after 4 freeze/thaw cycles [45.1 (4.6), n = 10] or overnight exposure to room temperature [45.7 (2.4), n = 6]. Compared with the 58 controls [peak size 50.2 (3.3)], the 2.740-kDa peak was lower in 17 patients with nonglioma brain pathologies [peak size 22.2 (3.0), P <10–7] and in 11 patients with colorectal cancer who had no brain metastases [peak size 30.0 (7.8), P <0.05]. Nonglioma brain pathologies were metastatic brain tumor (9), brain abscess (3), head injury (1), stroke (2), multiple sclerosis (1), and subarachnoid hemorrhage (1).

2.740-Kda peptide sequencing
The 2.740-kDa peptide was sequenced by Ciphergen Biosystems using tandem MS. SwissProt database search using Mascot revealed only 1 hit with Mowse Score 24 (TVVQPSVGAAAGPVVPPCPGRIRHFKW, P02765) corresponding to the short chain of human AHSG. In SELDI-TOF MS experiments, the serum was reduced so that the B-chain was released. The A-chain was undetectable with the CM10 surface used here. Because of the low Mowse score, we proved that the 2.740-kDa peak came from AHSG by immunodepletion. Nonreduced AHSG was immunodepleted from serum using a polyclonal anti-AHSG antibody coupled to protein G beads (Fig. 2A ). After AHSG depletion, the serum was reduced and processed as described in Materials and Methods. The 2.740-kDa peak selectively disappeared from the serum and appeared in the immunoprecipitated sample. When the anti-AHSG antibody was omitted, the 2.740-kDa peptide persisted in the serum spectrum and was absent from the immunoprecipitated fraction. Neither protein G nor antibody alone produced a 2.740-kDa peak. Purified, reduced human AHSG (Biovendor) gave a strong 2.740-kDa peak in SELDI-TOF MS.


Figure 2
View larger version (16K):
[in this window]
[in a new window]

  		Figure 2. Serum AHSG concentration.

(A), SELDI-TOF mass spectra of immunoprecipitation experiments. Serum (normal serum), Ab/PG-treated serum (serum depleted of AHSG by anti-AHSG antibody coupled to protein G), PG-treated serum (serum treated with protein G, omitting the anti-AHSG antibody), Ab/PG precipitate (anti-AHSG antibody/protein G immunoprecipitated serum fraction), PG precipitate (immunoprecipitated serum fraction using Protein G without the anti-AHSG antibody), AHSG (pure human AHSG protein). (B), Serum AHSG level in 47 controls, 36 grade II astrocytomas, 15 grade III astrocytomas, and 91 grade IV astrocytomas. Mean ± SE. Significant correlation between serum AHSG and tumor grade (r = –0.56, 95% CI –0.65 to –0.45, P <0.0001).

serum ahsg
We measured serum AHSG in all 142 astrocytoma patients and 47 of 58 control subjects using turbidimetry. (In the remaining 11 controls, there was insufficient serum to measure AHSG.) There was significant correlation (r = 0.76, 95% CI 0.61–0.86, P <0.0001) between the 2.740-kDa peak size and serum AHSG concentration. Serum AHSG concentrations progressively decreased with tumor progression from control [356.4 (6.3) mg/L] to grade II [254.9 (16.1) mg/L], grade III [239.7 (27.0) mg/L], and grade IV [202.5 (8.4) mg/L] astrocytoma (Fig. 2BUp ).

serum ahsg and survival of gbm patients
Serum AHSG was ≥285 mg/L in 100% of control subjects, but only in 39% of grade II, 33% of grade III, and 22% of grade IV astrocytoma patients. We used 285 mg/L as an arbitrary cutoff to divide the 91 GBM patients into 2 groups, 71 patients with AHSG <285 mg/L vs 20 with AHSG ≥285 mg/L. Kaplan-Meier survival curves for each group, beginning on the date of sample collection, are shown in Fig. 3A . GBM patients with low AHSG (<285 mg/L) had a median survival of 29 weeks vs 51 weeks for patients with normal AHSG (≥285 mg/L). Patients with higher serum AHSG were more likely to survive beyond a year (Fig. 3B ). The survival of GBM patients was worse in patients >60 years old (HR 1.9, 95% CI 1.3–3.5, P <0.005) or with Karnofsky score <70 (HR 2.3, 95% CI 1.7–4.8, P <0.0005). In a multivariate Cox proportional hazards model, serum AHSG <285 mg/L (HR 2.7, 95% CI 1.5–5.0, P <0.001), age >60 (HR 2.0, 95% CI 1.3–3.1, P <0.005), and Karnofsky score <70 (HR 2.5, 95% CI 1.6–4.0, P <0.0001) were all independent predictors of survival.


Figure 3
View larger version (17K):
[in this window]
[in a new window]

  		Figure 3. Serum AHSG predicts survival in GBM patients.

(A), Kaplan-Meier plots of GBM patients with low (<285 mg/L, n = 71) vs normal (≥285 mg/L, n = 20) serum AHSG. The 2 survival curves significantly differ (HR 1.9, 95% CI 1.2–3.2, P <0.01). (B), Serum AHSG vs 1-year survival of GBM patients. n = 6 (<100 mg/L), 36 (100–200), 38 (200–300), and 11 (>300 mg/L). (C, D), Kaplan-Meier plots of GBM patients grouped according to the number of favorable prognostic factors. In C, factors are age <60 years, Karnofsky score ≥70, and serum AHSG ≥285 mg/L (0, n = 14; 1, n = 32; 2, n = 39; 3, n = 6). In D, factors are age <60 years and Karnofsky score ≥70 (0, n = 18; 1, n = 35; 2, n = 38). Number of factors correlates with improved survival (P <0.0001 for C and D).

combined prognostic index
To illustrate the clinical potential of AHSG to predict survival in GBM patients, we combined patient age and Karnofsky score, the 2 most widely used prognostic factors(2)(3)(4), with serum AHSG into a simple 3-point score. Fig. 3CUp shows that median survival depends on the number of favorable preoperative prognostic factors (age ≤60 years, Karnofsky score ≥70, serum AHSG ≥285 mg/L). In contrast to a 2-point score using only age and Karnofsky score (Fig. 3DUp ), the 3-point score distinguishes a subgroup of GBM patients (group 3) with long (>2 years) median survival.

correlative studies
We explored 3 hypotheses to explain the correlation between survival and serum AHSG concentration in patients with GBM. First, we asked whether low serum AHSG concentration correlates with high comorbidity as previously suggested(16)(17), which predisposes patients to early death because of their inability to cope with the added burden of the tumor. Second, because AHSG was reported to behave as an inflammatory marker in some situations(18)(19), we wondered whether low serum AHSG concentration correlates with a deleterious inflammatory response to the tumor. Third, we investigated whether low AHSG concentration correlates with more malignant tumors.

We rejected the first hypothesis (low AHSG indicates high comorbidity) because there was no correlation between serum AHSG concentration and performance (Karnofsky, r = 0.04, P = 0.68) or comorbidity (Charlson, r = –0.13, P = 0.18) scores in the GBM patients. Although AHSG is produced by the liver(16)(18)(20) and may be involved in regulating serum calcium concentration(21)(22)(23), there was no correlation between AHSG concentration and serum calcium concentrations (r = 0.21, P = 0.28, n = 28) or alanine aminotransferase (r = –0.10, P = 0.50, n = 48) activities in GBM patients.

The second hypothesis, that low AHSG indicates deleterious inflammation, was also rejected, because no correlation was found between serum AHSG and hsCRP concentrations (r = 0.24, P = 0.10, n = 50) or between serum AHSG and blood leukocyte count (r = –0.16, P = 0.22, n = 59) in GBM patients. Serum hsCRP level did not differ between 50 controls vs 77 GBM patients [3.78 (0.72) vs 3.15 (0.70) mg/L, P = 0.53].

We accepted the third hypothesis—in GBM patients, serum AHSG level mirrored the degree of tumor malignancy. There was significant correlation between serum AHSG concentration and the percentage of Ki67 positive tumor cell nuclei (Fig. 4 ). MRS revealed marked correlation between serum AHSG and tumor lipid (r = –0.63, 95% CI –0.89 to –0.09, P <0.05, n = 12), myoinositol (r = 0.66, 95% CI 0.13–0.89,P <0.05), and N-acetyl-aspartate (r = 0.72, 95% CI 0.26–0.92, P <0.01) content (Fig. 5 ).


Figure 4
View larger version (34K):
[in this window]
[in a new window]

  		Figure 4. GBM proliferation inversely correlates with serum AHSG.

(A), Typical GBMs immunostained for Ki67 with corresponding serum AHSG in mg/L. Immunopositive nuclei are brown (DAB), and immunonegative nuclei are blue (hematoxylin). Bar = 40 µm. (B), Percent of Ki67-positive nuclei vs serum AHSG (mean ± SE) and best-fit trend line (r = –0.77, P <0.001, n = 14).


Figure 5
View larger version (58K):
[in this window]
[in a new window]

  		Figure 5. MR spectroscopy indices of GBM malignancy correlate with serum AHSG.

(A), (Left) MRI scan (T2-weighted) of a patient with GBM. The black square shows the region used to obtain the MR spectrum. (Right) mIG, myoinositol plus glycine; tCho, total choline; tCr, total creatine; Glx, glutamate plus glutamine; MM, macromolecules. (B), Lipid peak (1.3 ppm) vs serum AHSG and best-fit straight line (r = –0.63, 95% CI –0.89 to –0.09, P <0.05, n = 12). (C), Myoinositol peak vs serum AHSG and best-fit straight line (r = 0.66, 95% CI 0.13–0.89, P <0.05, n = 12).

ahsg validation
To confirm the predictive value of serum AHSG, we used a separate group of 72 patients with GBM collected from different hospitals and followed up at 30 weeks. Table 1Up summarizes patient demographics. There was no significant difference in survival between patients treated at St. George’s vs King’s College hospitals. Serum AHSG was ≥285 mg/L in 7 patients and <285 mg/L in 65 patients. In a multivariate Cox proportional hazards model, serum AHSG <285 mg/L was a significant prognostic factor (HR 6.4, 95% CI 1.1–6.6, P <0.05) independent of Karnofsky score and age. Patients with higher serum AHSG were more likely to be alive when followed up at 30 weeks (20.0% survived with AHSG <100 mg/L, 34.2% with AHSG 100–200, 39.1% with AHSG 200–300, and 83.3% with AHSG >300).


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
Our data suggest that preoperative serum AHSG is low in most patients with astrocytoma. About 10%–20% of GBM patients, however, have normal serum AHSG concentration (≥285 mg/L), and this is associated with prolonged survival, independent of patient age and Karnofsky score. The prognostic value of serum AHSG was validated in a separate group of GBM patients treated in 2 different centers. The functional significance of these findings is supported by the observation that low serum AHSG concentration correlates with increased tumor proliferation (high Ki67 immunolabeling index)(24)(25). The negative correlation of serum AHSG with tumor lipids and positive correlation with myoinositol suggests that serum AHSG inversely correlates with the fractional necrosis of GBM and hence its malignancy(12)(26).

We elected to use serum rather than cerebrospinal fluid (CSF) in our study because CSF collection is impractical. Preoperative lumbar puncture carries a high risk of lethal brain herniation in GBM patients because of brain edema and increased intracranial pressure. Perioperative collection of CSF is also impractical, because the CSF is often contaminated with blood from the surgery. Furthermore, it is unethical to collect CSF from normal subjects, whose inclusion in the study was essential. In contrast to the risks associated with CSF collection, serum is safely and easily obtainable with no risk to patients or normal subjects. A previous study reported increased AHSG in the CSF of patients with low-grade gliomas(27), who were found to have reduced serum AHSG in our study. AHSG immunohistochemistry (not shown) revealed AHSG leakage from the blood through disrupted blood–tumor barrier into the CSF in glioma patients, thus explaining the reported increase in CSF AHSG.

Reduced serum AHSG is not a specific biomarker of GBM. We showed that AHSG is also low in patients with non-glioma brain pathologies and in patients with non-CNS tumors. Sepsis, impaired liver function, and cardiovascular disease also reduce serum AHSG(17)(19)(20). No association was found between serum AHSG and liver function, comorbidity, or inflammation, which suggests that the mechanism of GBM-induced AHSG reduction is different. Lack of correlation between serum AHSG and hsCRP indicates that AHSG is not regulated as an acute-phase protein in GBM patients. The mechanism of reduced AHSG in GBM patients is therefore unknown, but may involve reduced AHSG production by hepatocytes or increased AHSG degradation by glioma cells. Although liver function blood tests in GBM patients were normal, a selective inhibition of hepatocyte AHSG production in glioma patients cannot be excluded. Elucidation of cell-level mechanisms will require cell culture studies to determine whether glioma cells signal hepatocytes to reduce AHSG production or whether glioma cells degrade AHSG.

Posttranslational modifications may alter the binding affinity of the 2.740-kDa peptide to the CM10 chip surface, as well as changing the peptide mass. In theory, therefore, the intensity of the 2.740-kDa peak does not necessarily indicate serum AHSG concentration. However, the strong positive linear correlation between 2.740-kDa peak intensity and total serum AHSG suggests that differences in 2.740-kDa peak size primarily represent differences in AHSG concentrations. It would be interesting to study whether posttranslational modifications of AHSG, including phosphorylation and glycosylation(28), carry additional prognostic information in GBM patients.

This is the first report that serum AHSG predicts survival in patients with GBM, especially when combined with patient age and Karnofsky score. The combined prognostic score suggests that two-thirds of young, independently functioning GBM patients with normal serum AHSG survive for >2 years, whereas >90% of old, dependent patients with low AHSG die within 6 months of surgery. Our prognostic score could prevent the morbidity caused by unnecessary adjuvant therapies in GBM patients with short life expectancy, and may also be helpful when counseling patients about disease progression and when evaluating novel treatments.

We are currently recruiting more patients to investigate whether serum AHSG also predicts outcome in patients with lower-grade astrocytomas. Interestingly, of the 15 patients with anaplastic astrocytoma, all who had normal serum AHSG (n = 5) survived >2 years, whereas those with low serum AHSG (n = 10) had a median survival of <10 months. However, the small number of grade II and grade III astrocytoma patients in this study prevents us from drawing definitive conclusions about the prognostic value of AHSG in these patients.

Several reports describe prognostic biomarkers in GBM tissue(13)(29)(30). Tissue biomarkers have disadvantages compared with serum biomarkers. The assays require a tumor specimen and therefore cannot be done preoperatively. Even when a tumor sample is available, tissue biomarkers may not be representative of a histologically heterogeneous tumor, as a biopsy samples only a small region. A key advantage of AHSG is that it can be easily measured preoperatively because AHSG is an abundant serum protein and ELISA kits are commercially available. AHSG is stable in serum for at least 2 weeks at 4 °C and after several freeze-thaw cycles, making AHSG measurement practical in clinical settings.


   Acknowledgments
 
Grant/Funding Support: Supported by the Wellcome Trust, Neurosciences Research Foundation, St. George’s NHS R&D, and the Harrison Fund.

Financial Disclosures: None declared.

Acknowledgments: We thank Marie-Catherine Le Bihan, Nathan Harris, and Vladimir Podust.


   Footnotes
 
1 Nonstandard abbreviations: GBM, glioblastoma multiforme; MS, mass spectrometry; AHSG, {alpha}2-Heremans-Schmid glycoprotein; MRS, magnetic resonance spectroscopy; CNS, central nervous system; hsCRP, high-sensitivity C-reactive protein; HR, hazard ratio; CSF, cerebrospinal fluid.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Smirniotopoulos JG. The new WHO classification of brain tumors. Neuroimaging Clin N Am 1999;9:595-613.[Web of Science][Medline] [Order article via Infotrieve]
  2. Laws ER, Parney IF, Huang W, Anderson F, Morris AM, Asher A, et al. Survival following surgery and prognostic factors for recently diagnosed malignant glioma: data from the Glioma Outcomes Project. J Neurosurg 2003;99:467-473.[Web of Science][Medline] [Order article via Infotrieve]
  3. Devaux BC, O’Fallon JR, Kelly PJ. Resection, biopsy, and survival in malignant glial neoplasms: a retrospective study of clinical parameters, therapy, and outcome. J Neurosurg 1993;78:767-775.[Web of Science][Medline] [Order article via Infotrieve]
  4. Lacroix M, Abi-Said D, Fourney DR, Gokaslan ZL, Shi W, DeMonte F, et al. A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival. J Neurosurg 2001;95:190-198.[Web of Science][Medline] [Order article via Infotrieve]
  5. Cervi D, Yip TT, Bhattacharya N, Podust VN, Peterson J, Abou-Slaybi A, et al. Platelet-associated PF-4 as a biomarker of early tumor growth. Blood 2007;111:1201-1207.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  6. Kanmura S, Uto H, Kusumoto K, Ishida Y, Hasuike S, Nagata K, et al. Early diagnostic potential for hepatocellular carcinoma using the SELDI ProteinChip system. Hepatology 2007;45:948-956.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  7. Papadopoulos MC, Abel PM, Agranoff D, Stich A, Tarelli E, Bell BA, et al. A novel and accurate diagnostic test for human African trypanosomiasis. Lancet 2004;363:1358-1363.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  8. Hortin GL. The MALDI-TOF mass spectrometric view of the plasma proteome and peptidome. Clin Chem 2006;52:1223-1237.[Abstract/Free Full Text]
  9. Karnofsky DA, Abelman WH, Craver LF, Burchenal JH. The use of nitrogen mustards in palliative treatment of carcinoma. Cancer 1948;1:634-656.[CrossRef][Web of Science]
  10. Charlson ME, Pompei P, Ales KL, McKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-383.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  11. Opstad KS, Murphy MM, Wilkins PR, Bell BA, Griffiths JR, Howe FA. Differentiation of metastases from high-grade gliomas using short echo time 1H spectroscopy. J Magn Reson Imaging 2004;20:187-192.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  12. Howe FA, Barton SJ, Cudlip SA, Stubbs M, Saunders DE, Murphy M, et al. Metabolic profiles of human brain tumors using quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 2003;49:223-232.[CrossRef][Medline] [Order article via Infotrieve]
  13. Ohgaki H, Kleihues P. Population-based studies on incidence, survival rates, and genetic alterations in astrocytic and oligodendroglial gliomas. J Neuropathol Exp Neurol 2005;64:479-489.[Web of Science][Medline] [Order article via Infotrieve]
  14. Chang SM, Parney IF, Huang W, Anderson FA, Jr, Asher AL, Bernstein M, et al. Patterns of care for adults with newly diagnosed malignant glioma. JAMA 2005;293:557-564.[Abstract/Free Full Text]
  15. Woltje M, Tschoke B, von Bulow V, Westenfeld R, Denecke B, Graber S, et al. CCAAT enhancer binding protein {beta} and hepatocyte nuclear factor 3{beta} are necessary and sufficient to mediate dexamethasone-induced up-regulation of alpha2HS-glycoprotein/fetuin-A gene expression. J Mol Endocrinol 2006;36:261-277.[Abstract/Free Full Text]
  16. Honda H, Qureshi AR, Heimburger O, Barany P, Wang K, Pecoits-Filho R, et al. Serum albumin, C-reactive protein, interleukin 6, and fetuin a as predictors of malnutrition, cardiovascular disease, and mortality in patients with ESRD. Am J Kidney Dis 2006;47:139-148.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  17. Stenvinkel P, Wang K, Qureshi AR, Axelsson J, Pecoits-Filho R, Gao P, et al. Low fetuin-A levels are associated with cardiovascular death: impact of variations in the gene encoding fetuin. Kidney Int 2005;67:2383-2392.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  18. Lebreton JP, Joisel F, Raoult JP, Lannuzel B, Rogez JP, Humbert G. Serum concentration of human alpha 2 HS glycoprotein during the inflammatory process: evidence that alpha 2 HS glycoprotein is a negative acute-phase reactant. J Clin Invest 1979;64:1118-1129.[Web of Science][Medline] [Order article via Infotrieve]
  19. Ketteler M, Bongartz P, Westenfeld R, Wildberger JE, Mahnken AH, Bohm R, et al. Association of low fetuin-A (AHSG) concentrations in serum with cardiovascular mortality in patients on dialysis: a cross-sectional study. Lancet 2003;361:827-833.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  20. Kalabay L, Jakab L, Prohaszka Z, Fust G, Benko Z, Telegdy L, et al. Human fetuin/alpha2HS-glycoprotein level as a novel indicator of liver cell function and short-term mortality in patients with liver cirrhosis and liver cancer. Eur J Gastroenterol Hepatol 2002;14:389-394.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  21. Swallow CJ, Partridge EA, Macmillan JC, Tajirian T, DiGuglielmo GM, Hay K, et al. alpha2HS-glycoprotein, an antagonist of transforming growth factor beta in vivo, inhibits intestinal tumor progression. Cancer Res 2004;64:6402-6409.[Abstract/Free Full Text]
  22. Schafer C, Heiss A, Schwarz A, Westenfeld R, Ketteler M, Floege J, et al. The serum protein alpha 2-Heremans-Schmid glycoprotein/fetuin-A is a systemically acting inhibitor of ectopic calcification. J Clin Invest 2003;112:357-366.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  23. Hermans MM, Brandenburg V, Ketteler M, Kooman JP, Sande FM, Gladziwa U, et al. Study on the relationship of serum fetuin-A concentration with aortic stiffness in patients on dialysis. Nephrol Dial Transplant 2006;21:1293-1299.[Abstract/Free Full Text]
  24. Scott IS, Morris LS, Rushbrook SM, Bird K, Vowler SL, Burnet NG, et al. Immunohistochemical estimation of cell cycle entry and phase distribution in astrocytomas: applications in diagnostic neuropathology. Neuropathol Appl Neurobiol 2005;31:455-466.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  25. Torp SH. Diagnostic and prognostic role of Ki67 immunostaining in human astrocytomas using four different antibodies. Clin Neuropathol 2002;21:252-257.[Web of Science][Medline] [Order article via Infotrieve]
  26. Gruber S, Stadlbauer A, Mlynarik V, Gatterbauer B, Roessler K, Moser E. Proton magnetic resonance spectroscopic imaging in brain tumor diagnosis. Neurosurg Clin N Am 2005;16:101-114, vi.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  27. Ribom D, Westman-Brinkmalm A, Smits A, Davidsson P. Elevated levels of alpha-2-Heremans-Schmid glycoprotein in CSF of patients with low-grade gliomas. Tumour Biol 2003;24:94-99.[CrossRef][Medline] [Order article via Infotrieve]
  28. Gejyo F, Chang JL, Burgi W, Schmid K, Offner GD, Troxler RF, et al. Characterization of the B-chain of human plasma alpha2HS-glycoprotein. The complete amino acid sequence and primary structure of its heteroglycan. J Biol Chem 1983;258:4966-4971.[Abstract/Free Full Text]
  29. Hill C, Hunter SB, Brat DJ. Genetic markers in glioblastoma: prognostic significance and future therapeutic implications. Adv Anat Pathol 2003;10:212-217.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  30. Hakin-Smith V, Jellinek DA, Levy D, Carroll T, Teo M, Timperley WR, et al. Alternative lengthening of telomeres and survival in patients with glioblastoma multiforme. Lancet 2003;361:836-838.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
clinchem.2007.096792v1
54/4/713    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Petrik, V.
Right arrow Articles by Papadopoulos, M. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Petrik, V.
Right arrow Articles by Papadopoulos, M. C.
Related Collections
Right arrow Proteomics and Protein Markers

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS