Predictive value of tumor recurrence using urinary vascular endothelial factor levels in patients receiving radiation therapy for Glioblastoma Multiforme (GBM)

Glioblastoma Multiforme (GBM) is the most frequent primary malignant brain tumor in adults [1, 2]. Its treatment generally includes maximally safe surgical resection followed by adjuvant radiation therapy and temozolomide chemotherapy [3]. Despite advances in understanding the biology behind GBM [1, 4, 5] and some improvement in outcome with the addition of temozolomide chemotherapy, its prognosis remains grim with a median overall survival (OS) of 15 months [6].

Personalized therapy based on tumor biology in individual patients to improve treatment outcome has identified several potential biomarkers [5, 7, 8], including vascular endothelial growth factor (VEGF) [9]. Increased VEGF activity has been associated with early recurrence in patients with cancer including GBM [10–12]. VEGF can be measured in biofluids such as urine, serum or plasma [13–15]. The measurement of VEGF in urine is an attractive option as urine collections are noninvasive, there are minimal preparation costs and urinary proteins are stable during long term storage [16, 17]. In addition, the measurement of VEGF in serum or plasma can be problematic as VEGF can be released by platelets in the process of phlebotomy in the case of serum [15] and sequestration of VEGF within platelets may make accurate VEGF measurement difficult in the case of plasma [13, 14]. VEGF as a urinary biomarker has previously been shown to be a predictive marker for progression free survival (PFS) in cancer patients after the completion of radiotherapy (RT) [10]. Its potential importance to the treatment of GBM is based on its role in pro-angiogenesis.

VEGF is responsible for increased vascularity and vascular permeability in tumors as well as migration and proliferation of endothelial cells [18, 19]. VEGF mRNA has been identified in pseudopalisading cells around necrotic zones in GBM and its expression correlates with grade in diffuse astrocytomas [20]. We hypothesized that increased urinary VEGF levels would predict for recurrence or progression in patients with GBM patients. The purpose of this study was to examine the feasibility of collecting and analyzing urine in a prospective, multi-institutional trial and the potential utility of urinary VEGF as a predictive biomarker for progression. In order to explore the dynamic nature of this potential relationship, VEGF was measured in the urine of GBM patients prior to radiation therapy, at completion and at 1 month follow-up.

At 1 year, 35 (27.3%) patients had not progressed, 89 (69.5%) had progressed, and 4 (3.1%) died without progression. For the entire group of 128 patients, at baseline the median VEGF level was 52.9 pg/mgCr (range 0.2-15,034.4); on the last day of RT 56.6 pg/mgCr (range 0 – 2,377.1); and at 1 month following RT 70.0 pg/mgCr (range 0.1-1,813.2). In patients who had not progressed at 1 year, both their baseline VEGF levels of 40.3 pg/mgCr (range 0.2-350.8) as well as their end of RT VEGF levels of 41.8 pg/mgCr (range 0–356.8) were lower than the levels in patients who had progressed at one year, 59.7 pg/mgCr (range 1.3-15,034.4) and 69.7 pg/mgCr (range 0–2,377.1), respectively. However, the differences among these groups did not reach statistical significance.

The normalized VEGF change from the end of RT to the 1 month post-RT measurement was compared between patients who progressed by 1-year vs. those that did not progress (Table 2). Chi-square test revealed no significantly higher rate of failure in patients with elevated VEGF at 1 month radiation therapy (p = 0.21). On univariate and multivariate logistic regression analysis no significant correlations were identified for any of the variables examined (randomization arm on RTOG 0525, methylation status, RPA class, baseline normalized VEGF, and VEGF change from end of radiation therapy to 1 month follow-up) (Table 3). The area under the Receiver Operating Characteristic (ROC) curve (AUC) is 0.5401, suggesting that VEGF is not a useful biomarker to predict recurrence at 1 year (Figure 1).

VEGF has emerged as a significant drug target due to its importance as a pro-angiogenic cytokine in gliomas [4]. VEGF signaling is thought to occur in response to hypoxia and results in the formation of vessels whether de novo or from preexisting vasculature [25]. Its role may be especially important in gliomas as they exhibit aberrant vessels, hypoxia and endothelial proliferation. Increasing VEGF activity correlates with an increase in vascular proliferation and increasing tumor grade in gliomas [26]. A number of drugs aimed at neutralizing VEGF receptors are currently being investigated [9], most notably bevacizumab, a monoclonal antibody that binds and neutralizes VEGF. Bevacizumab has been approved for use in non-small cell lung cancer, metastatic renal cell carcinoma, and based on the BRAIN [27] and the NCI 06-C-0064E studies [28] it has also been approved for use in patients with recurrent GBM. The two GBM trials showed a reduction in steroid use, temporary improvement in neurological function and, in the BRAIN study, a PFS of 9.2 months. In the newly diagnosed GBM setting, a phase II trial [29] has shown a PFS benefit with no OS benefit when Avastin is used in conjunction with RT and temozolomide, although subset analysis showed that RPA class IV/V may have an OS benefit as well. The addition of Avastin to the standard regimen of radiation and temozolomide is currently being investigated in large randomized phase III trials (RTOG 0825, BO21990).

Previously, Chan et al. [10] showed that urinary VEGF levels at the end of radiotherapy compared to the level at least 1 month post treatment to be predictive of 1 year PFS in a variety of tumor types, including GBM. Thus, the present study focused on VEGF as a potential biomarker for tumor recurrence in patients with GBM receiving chemoradiation. Our results show that the patient population in the current study is largely representative of the RTOG0525 population, suggesting applicability to GBM patients in general. However, we measured no significant difference with respect to changes in VEGF in the patients who failed treatment and those who did not. Furthermore, no significant correlation was found on univariate/multivariate analysis for either baseline VEGF or the change in VEGF level. However only 25% of patients were found to have methylated MGMT in the current study as compared to historical rates on the order of 45% [5] to 50% [30–32]. As methylated MGMT has been shown to be both prognostic and predictive for an improved outcome in GBM [33–36], this may affect the survival outcome as patients may be more likely to fail treatment. Recent analysis of the RTOG0525 results does not suggest that this is the case as their outcomes are comparable to published literature [31]. Randomization to either arm was not statistically different suggesting that standard chemotherapy vs. dose dense chemotherapy is likely not a confounding factor with respect to urinary VEGF (Table 1).

Limitations of the study include the small number of patients (128) for whom adequate samples were available as well as potentially the timing of VEGF measurement and the lack of long term follow-up of urinary VEGF levels. Although previous data [10] showed a correlation between urinary VEGF levels and PFS in a population of cancer patients, there is the possibility that the correlation with respect to GBM may exist but is simply not being detected in the current study as result of the timing of VEGF measurement. VEGF secretion in quantities large enough to be predictive for recurrence may not occur until a significant burden of disease is present, which is less likely at the end of treatment or even at the 1 month follow-up point. The possibility remains that VEGF levels could in fact be predictive at a later date that is still before a definitive radiographic diagnosis could be made. In such a scenario, they could potentially help distinguish progression from pseudoprogression.

Despite this lack of statistical significance, baseline values for VEGF were higher for patients who progressed by 1 year. This was the case both at baseline as well as at the end of radiotherapy. This may support the idea that at the very least urinary VEGF may have value as a prognostic if not necessarily as a predictive biomarker in GBM.

Recent studies have revealed significant heterogeneity within gliomas and even within the narrowly defined GBM histology [1, 37, 38]. Based on the GBM subtypes identified in Verhaak et al., it is possible that VEGF levels may be predictive in some GBM subtypes specifically the mesenchymal or classical. EGFR activation (classical subtype) and PTEN mutation (mesenchymal subtype) have both been associated with increased VEGF expression [4]. Further studies could examine a possible correlation between VEGF levels and molecular GBM subtypes and possibly include VEGF level within the risk classification itself. A possible correlation between GBM subtypes and VEGF was not addressed in the current study. RTOG 0825, one of the two ongoing phase III trials involving bevacizumab, includes patient stratification by mesenchymal/angiogenic gene expression and may indirectly explore this potential relationship.

Urinary VEGF was not identified as a predictive factor for tumor recurrence in GBM patients receiving radiation therapy. There is a suggestion that baseline and post treatment VEGF levels may be higher in patients who ultimately fail treatment although this was not statistically significant. This study confirmed that urine can be collected and analyzed in a prospective, multi-institutional trial.

Written informed consent was obtained from the patients for the publication of this report.