Introduction

This report describes a further piece of information considering the use of valproate (valproic acid, VPA) as novel caner drug in pediatric patients. VPA (di-n-propylacetic acid, MW 144.21) is a new anticancer agent, but the drug has been used for treatment of other diseases. It was first used as antiepileptic drug [1] and mood stabilizer [2], with very-well documented pharmacokinetic and toxicity profiles [3, 4]. Desired trough serum levels for antiepileptic efficacy range between 50–100 mg/l (equivalent to 0.3–0.6 mM). At therapeutic levels, the drug is teratogenic [5]. At higher drug doses, toxic effects such as thrombocytopenia, pancreatitis, hypothyroidism, and polycystic ovarian syndrome [6] are common side effects. In parallel to its antiepileptic mode of action, VPA was found to induce differentiation in the neuroectodermal [7] and hematopoetic systems [8, 9], and even an anti-tumor response was described in vitro [10, 11] and in patients [12, 13]. As an underlying mechanism for this observation inhibition of histone deacetylase by VPA was identified by Phiel [14]. Histone acetylation is a major mechanism of epigenetic regulation, loosening the DNA packing. Histone deacetylase down-regulates gene expression by wrapping the DNA tighter around the histone proteins, thus making genes less accessible to promoter binding proteins. Histone-deacetylase inhibition results in more acetylase groups binding to the histones and therefore promotes gene expression. Since this finding, a large body of literature describing details of VPA’s mechanism has been produced [15, 16] confirming that VPA induces tumor cell differentiation and inhibits proliferation as well as it increases the rate of apoptosis [17] and immunogenicity [18]. P21/WAF activation independent of p53 activation is one of the mechanisms of action [19] as well as ERalpha degradation [20] and inactivation of NF-KappaB [21], but the complete picture of intracellular events is far more complex including acetylation of p53 directly [22]. Furthermore, these effects are accompanied by the reduction of the tumor cells’ metastatic and angiogenic potency. Newly designed drugs with more specific functions, inhibiting certain types of histone deacetylases, have also been developed [18, 23–25]. As studies with histone deacetylase inhibitors move from preclinical models [26] to clinical trials [27], and studies in adult patients have shown tolerability [28], we asked if the drug can be given safely to heavily pretreated pediatric patients with high grade glioma and diffuse intrinsic pontine glioma.

Pediatric high-grade glioma (HGG) and diffuse intrinsic pontine glioma (DIPG) remain challenging diseases. Progress in developing treatments is slow because the diseases are rare and therefore are frequently combined in clinical studies [29, 30–34]. Radiation therapy is generally accepted as beneficial for both disease groups. Chemotherapy has been shown to be beneficial in children [17, 35] and in adults [36]; however, the overall results leave much room for improvement.

The HIT-GBM® study (acronym for “Hirntumor—glioblastoma multiforme”) is a prospective cohort-comparison study for this patient population [37] and a prospective registry part for control patients [29, 38, 39] following other treatment approaches. To elucidate whether Valproic acid (VPA) treatment contributes to long term survival in this pediatric patient population, VPA was added to the treatment protocol at the end of the intensive radiochemotherapy treatment protocol HIT-GBM-C as a rear-window. In addition, VPA was used as relapse treatment in other patients, who had relapsed on the intensive chemotherapy treatment before reaching the prescribed time to start VPA. This report describes our findings in both patient groups.

Material and methods

Treatment protocol

Valproic acid treatment was a part of the complex HIT-GBM-C treatment protocol. Patients were eligible if they were between 3 and 18 years with newly diagnosed DIPG, or grade III or IV HGG. Eligibility required an informed consent approved by the Ethics Committee, as required by rules and regulations at that time and in accordance with the 1964 Declaration of Helsinki. Treatment started with standard fractionated focal radiation and simultaneous chemotherapy. The first chemotherapy cycle was the PEV cycle [40, 41], which consists of cisplatin, 20 mg/m2/day on days 1–5; etoposide, 100 mg/m2 on days 1–3; and vincristine, 1.5 mg/m2 only on day 5. During radiation, weekly vincristine injections were administered until the last week of radiation therapy, in which patients were given one cycle of PEIV [38, 41], combining PEV with ifosfamide, 1.5 g/m2/day on days 1–5. After simultaneous radiation-chemotherapy, chemotherapy continued with further cycles of PEIV in weeks 10, 14, 18, 22, 26, and 30, as well as vincristine in weeks 13, 17, 21, 25, and 29, using the same doses. Upon completion of eight intensive HIT-GBM-C chemotherapy cycles, VPA was given as additional maintenance treatment. If the tumor was stable by MRI criteria after the first two cycles of HIT-GBM-C chemotherapy, two more cycles of chemotherapy were recommended, but if the disease remained stable at cycle 4 or beyond, further chemotherapy cycles were skipped to start VPA treatment earlier. Also, in case of tumor progression during the chemotherapy, patients were offered valproic acid as relapse treatment.

VPA was given orally as a single agent, starting with 10 mg/kg/day in week 1 and 20 mg/kg/day in week 2. Serum trough levels were then analyzed and the dosage adjusted to reach target serum levels of 100–150 mg/l, unless unwanted side effects had already occurred. Before and after 6 weeks of VPA treatment, the tumor was evaluated by MRI. Tumor response was defined by one-dimensional measures on MRI images. Progressive disease was defined as 25% or more tumor growth. Partial response was defined as 50% or more tumor shrinkage. Measurement in between were defined as stable disease.

Statistics

In this report we extracted data from the HIT-GBM database, which documents 310 patients diagnosed with HGG or DIPG between 1995 and 2003. This population was recently described in detail [42]. To avoid selection bias and describe the real world of this patient population, only patients receiving VPA for epilepsy during chemotherapy were excluded, while patients treated with major violations during the chemotherapy protocol were not excluded. The analysis includes patients following the original HIT-GBM-C protocol and patients originally only registered as control patients but then treated following the relapse protocol with VPA. Patients who were treated following the protocol but did not reach therapeutic valproic acid level were not excluded. The results were analyzed using the statistical package SPSS 12.0 (SPSS Inc., Chicago, Illinois, USA).

Results

Patients

Forty-four of 310 HIT-GBM database [42] patients received valproic acid for tumor treatment. The epidemiologic characteristics were as expected from this patient population; details are described in Table 1.

Table 1 Patient population
Full size table

Toxicity and serum levels

Despite the heavily pretreated patient population, the toxicity of VPA was very mild. There was no toxic death, no SAE was reported. Adverse events only required reporting in case of grade III or IV. No such event was reported. However, minor toxic reactions were reported in few patients (Grade I or II). These included: somnolence, three patients; difficulties in concentrating, one; enuresis, one; increased sensitivity to light, one; thrombocytopenia, one; and anemia, one. Serum trough levels were reported in 23 patients. The mean level of 99 mg/l (standard deviation: 26 mg/l) was well in keeping with the target level. The range of 52–175 mg/l was wide, and toxicity was reported more frequently in the patients with lower drug levels. There was no detectable relationship between VPA level and outcome.

Outcome

Tumor status was evaluated after 6 weeks of VPA treatment. Of the 22 patients who started treatment according to protocol, 19 remained stable. Progressive disease occurred in 1 of 12 patients who started VPA therapy with no measurable tumor and 2 of ten patients who started VPA treatment with stable measurable tumors. Of the 22 patients who received VPA as relapse treatment, 14 had MRI examinations at the required time points before and 6 weeks after starting VPA. Of those one responded (PR), seven had stable disease, and six had further progression of their disease.

The overall median survival for all 44 patients after starting valproic acid was 1.33 years (SE, 0.41; 95% confidence interval, 0.98–3.60 years). The survival time largely depended on the tumor status when treatment was initiated, with the data confirming what appears intuitive: patients who started the treatment without documented tumor progression had better results (Fig. 1). Median survival durations were 4.05 years (95% CI, 2.91–5.19 years) for these patients, compared with 0.58 years (95% CI, 0.13–1.02) for patients who started valproic acid after relapse, and the 5-year overall survival rates were, respectively, 44% ± 12.7% SE compared with 14% ± 7% (P = 0.0013, log–rank test). However, even among patients who started valproate with a progressive tumor, there were three long-term survivors alive after 4.5, 4.95, and 5 years. Gender had no influence on survival (P = 0.7); the influences of tumor location [42], extent of resection [43], presence of metastases [44], and histologic type [42] were similar to those of previous analyses but because of small numbers were not significant.

Fig. 1
figure 1

Overall survival of pediatric patients with HGG or DIPG treated with valproic acid. Twenty-two patients started treatment as maintenance treatment in the protocol HIT-GBM-C without signs of tumor progression (black line), and 22 patients started Valproate as relapse treatment (gray line). Data are classified as censored, when the patient is alive at the last date of information

Full size image

Discussion

Our analysis of the data showed that valproic acid treatment was well tolerated in 44 heavily pretreated pediatric HGG and DIPG patients when given orally as continuous treatment with individual dose adaptation. This is an important finding and will allow building this way to give valproate into multi-agent treatment protocols.

Valproic acid is one of the few drugs for which doses are frequently modulated according to drug levels in the blood. However, neither the tumor effect nor the side effects are closely related to drug levels in this study. In a reported case series of severe valproic acid intoxications [45], patients with peak valproic-acid concentrations above 450 mg/l were more likely to develop significant clinical effects and have longer hospital stays than were patients with lower concentrations. In that same study, peak valproic acid concentration above 850 mg/l was more likely than were lower concentrations to be associated with coma, respiratory depression, aspiration, or metabolic acidosis.

In adult patients intravenous valproate given over 5 days reached a maximal tolerated dose of 60 mg/kg/day with neurotoxicity being the dose limiting toxicity [28]. Those doses appear far higher, but they were only given for 5 days, resulting in fact in a lower total cumulative dose than given here. The comparison with our experience shows that the dose limiting toxicity was dependent of the route or schedule how the drug was given. In pediatric patients a feared rare side effect of antiepileptic VPA treatment is hepatopathia, which has an incidence of 1:5000 in the pediatric population. Worldwide, 250 fatal cases have been described, as reported by Koenig and colleagues [46]. In that study, the most recently described cases had a median drug level of only 78 mg/l. Hepatopathia is associated with increased liver enzyme (alanine transaminase (ALT), aspartate transaminase (AST), and gamma-glutamyl transpeptidase (gamma-GT)) levels in blood, decreased levels of fibrinogen, and increased levels of ammonia. Pre-existing metabolic diseases and co-medication with other antiepileptic drugs were frequent among these children [46]. Discontinuation of valproic acid and supportive care, including intravenous carnitine [47], is the recommended treatment if severe toxic reactions occur. In the 44 patients reviewed in our study, such toxic reactions did not occur despite the intensive chemotherapy that these children had prior to starting Valproate. Possibly the prior intensive chemotherapy functioned as a filter, preventing children with predispositioning disease from ever starting valproic acid treatment: Nevertheless, these data should not lead to a neglect of metabolic diagnostics before starting Valproate, particularly in fist-line treatment. The inverse relationship between drug levels and reported toxicity noted in our patients was caused by the dose recommendation of the study: doses were only increased in the absence of side effects. The data show that a wide range of inter-individual tolerance to the drug exists and that dosing by individual tolerability allows higher doses in some patients than does dosing by body weight or body surface area or drug levels.

When considering tumor efficacy one needs to acknowledge the particular poor prognostic patient group involved in this study (Table 1). Among the patients who received Valproate as the last part of the HIT-GBM-C protocol, the drug did not shrink any of the remaining heavily pretreated tumors. Still, findings of only three early progressive tumors among 22 patients and a projected 5-year overall survival of more than 40% are quite promising in this patient population [42]. This is confirmed by findings in our second subset of patients. One of these relapse patients responded. This patient had been described in details before [48]. We had also described another patient previously with recurrent HGG who responded to Valproate treatment [12], who was not a part of this study. Furthermore, one of us (P–H. D.) is aware of other patients with response to VPA outside of this data set, which are to be reported separately. However, even the fact that seven of the 14 patients prospectively enrolled in the database had stabilization of disease, and that the survival curve (Fig. 1) is strongly suggestive of a positive effect, and in particular that there were three long-term surviving patients confirm the impression that valproic acid has a stabilizing effect on pediatric HGG.

Still, the benefit these patients had from valproic acid was not overwhelming, and the golden key to cure this disease was not yet found by using VPA as a single agent. The future utility of the drug might be in combination with classic chemotherapeutic agents or radiation. Synergism was shown in preclinical models [15, 49–51], and preliminary clinical data suggests that the chemotherapy toxicity is not largely increased when Valproate is given as an antiepileptic during chemotherapy [52].

In conclusion, we found that valproic acid as a tumor treatment for pediatric patients with HGG and DIPG is safe when dosed by individual tolerance. Evidence for efficacy is provided. We recommend further clinical studies combining the molecule with chemotherapeutic agents, radiation, or both.