Introduction

Malignant gliomas are the most common primary neoplasms of the human nervous system. Although the result of radiotherapy (RT) is greater with concurrent and adjuvant temozolomide (TMZ), median overall survival remains at approximately 12 months in patients with glioblastoma, the most aggressive form of glioma. Recently, studies of pseudoprogression (psPD), defined as spontaneous improvement of enhancing lesions without further treatment other than adjuvant TMZ, have been reported in patients with malignant gliomas treated with chemoradiotherapy along with TMZ [1–3]. The reported incidence ranged from 15 to 30%, and some studies demonstrated that the development of psPD is associated with favorable overall survival [3, 4]. Thus, if possible, differential diagnosis of early tumor responses, such as psPD and real progression, can assist physicians in identifying patients who may benefit from intensive salvage treatment. The methylation of O6-methylguanine DNA methyltransferase (MGMT) promoter is the only predictive molecular alteration for the development of psPD in patients with malignant gliomas treated with RT and maintenance TMZ to date. The further effort to identify the other useful predictive maker(s) is clearly required.

TP53 is a tumor suppressor gene commonly found in human cancers. p53 expression is induced by many types of stressors such as heat, cold, low-dose ionizing radiation, and low pH [5]. It is widely assumed that normal p53 functions play a crucial role in mediating cell death following DNA-damaging therapies. There is evidence to support the premise that wild-type p53 suppresses the suicide DNA repair enzyme MGMT, resulting in a higher efficacy of TMZ in p53 wild-type than p53 mutant tumors [6, 7]. Conversely, one study reported the increased sensitivity of mutant p53 glioma cells toward ultraviolet (UV)-C light by DNA damage-induced blockage of RNA polymerase II and transcription [8]. Therefore, p53 expression may be associated with chemotherapy or radiation sensitivity and subsequent development of psPD.

In the present study, we examined the incidence of psPD and correlated the development of psPD with clinical outcomes in patients with malignant gliomas undergoing treatment with concurrent chemoradiotherapy and TMZ. Further analysis was directed toward identifying clinical and molecular parameters associated with the development of psPD, including p53 overexpression.

Patients and methods

Study population

Between June 2004 and February 2009, a total of 40 malignant glioma patients treated with chemoradiotherapy and TMZ in the Seoul National University Bundang Hospital were reviewed. Five patients were excluded due to indeterminate biopsy results in four patients and loss of follow-up in one patient. Data for 35 patients were available and used for analysis. Clinical records were reviewed concerning performance status, the extent of surgery, histology, corticosteroid dose, neurologic signs and symptoms, and survival. The median duration of follow-up was 13.2 months (range 3.3–58.5 months) and the median follow-up duration of the survivors was 14.2 months (range 4.5–58.5 months).

Treatment details

Radiation therapy was delivered 5 days per week using a continuous course of treatment with photons produced with a megavoltage linear accelerator. The total radiation dose ranged from 59.4 to 61.2 Gy, using a conventional fractionation schedule of 1.8–2 Gy per fraction per day except in one patient whose tumor was located adjacent to the brainstem and who received a dose of 55.8 Gy. The radiation fields encompassed the residual tumor volume and tumor bed (as defined by a CT or MRI scan) with a 2-cm margin. Eight patients received fractionated stereotactic RT (FSRT) with intensity modulation with reduced margin after 40–50 Gy of irradiation. Concomitant TMZ (75 mg/m2/day) was given on all days during RT and was followed after 4 weeks by adjuvant TMZ (150–200 mg/m2/day) given on Days 1–5, every 28 days. The median number of cycles of adjuvant TMZ was 4 (range 1–15 cycles), and fourteen patients completed planned adjuvant treatment.

Diagnosis of pseudoprogression

Early evaluation of treatment response was initiated 4 weeks after the end of concomitant chemoradiotherapy (CCRT) with MRI. The size of enhancing lesion in the gadolinium-enhanced T1 weighted image was measured, and tumor progression was defined according to MacDonald’s Criteria [9] as a 25% increase in tumor size with two-dimensional measurement, newly appeared lesions, an increased need for corticosteroids, or neurological deterioration. Early progression (ePD) was defined as progression on the first and second MRI scan after CCRT and adjuvant TMZ was suspended at second MRI scan. If the progressive lesion decreased in size or stabilized on subsequent follow-up MRI without additional treatment, it was defined as psPD. Since majority of lesions were not spherical, we performed serial volumetric analysis of enhancing lesions to calculate a change of enhancing volume using RT planning software (BrainLAB, Feldkirchen, Germany). In this case progression was defined as a 40% increase in volume, which is in accordance with previous work [10]. If patients presented stable disease or had no lesion at the first evaluation after CCRT, additional cycles of TMZ were continued, and these were defined as the nonPD subgroup.

Immunohistochemistry of p53

The expression of p53 in tumor samples was determined by immunohistochemistry using the p53 monoclonal mouse antihuman antibody DO-7 (DAKO, Glostrup, Denmark). Four-micrometer-thick tissue sections were taken from formalin-fixed, paraffin-embedded, archival tumor blocks. For antigen retrieval, tissue sections on silanized slides (DAKO) were boiled in 0.01 M citrate buffer (pH 6.0, SIGMA, St. Louis, MO) for 15 min using a microwave. A commercially available kit (ChemMate DAKO EnVision Detection Kit, Peroxidase/DAB, Rabbit/Mouse, DAKO) was used, and the slides were developed with DAB (diaminobenzidine tetrahydrochloride) and counterstained with Harris-hematoxylin (DAKO). Positively-stained tumor cells were counted under a microscope. Expression of p53 in over 10% of the tumor cells was considered as overexpression, which is highly suggestive of p53 mutation, because the wild-type protein is not generally detectable due to its short half-life [11–13].

Statistical analysis

Statistical analysis was performed using SPSS version 17.0 (SPSS, Chicago, IL). The Kaplan–Meier method was employed to calculate the time to progression (TTP) and overall survival (OS). We designated the TTP as the time period from the date of surgery to the point when radiologic or clinical evidence of PD was noted and the OS as the interval between the date of surgery and the date of death. The difference in survival between different subgroups (nonPD vs. ePD vs. psPD) was assessed using the log-rank test. To define the association between clinical or biological factors and the development of psPD, the Fisher’s exact test was used, comparing psPD versus ePD. Differences were considered statistically significant at P < 0.05 (two-sided test).

Results

Patient characteristics

The patient population was comprised of seventeen men and eighteen women aged 23–82 years (median: 57 years). The histologic diagnosis was anaplastic oligodendroglioma in two patients, anaplastic astrocytoma in eight patients, and glioblastoma in twenty-five patients. Cytoreductive surgery was performed in twenty-two patients. Radical resection was accomplished in thirteen patients (59%), nine patients (41%) underwent subtotal resection, and the remaining eight patients were diagnosed by biopsy (Table 1).

Table 1 Patients characteristics
Full size table

Incidence of psPD

An early increase in the size of enhancing lesions was observed in eighteen patients at the first MRI scan after CCRT. Eight patients were classified as having psPD (23%), and the remaining ten patients were considered to have ePD (29%) after a second MRI scan. Among the patients with ePD, half had increased enhancing lesions and the others had a clinically or neurologically progressed condition. For volumetric measurement, only two patients had changes in their MRI volumetric response compared with diametric MacDonald’s Criteria: one patient went from ePD to psPD and the other from ePD to nonPD. The volumetric changes in enhancing lesions were demonstrated in Fig. 1a (psPD) and Fig. 1b (ePD).

Fig. 1
figure 1

The volume changes of enhancing lesions as a fraction of pre- or post-radiotherapy MRI volume are shown for patients with psPD (a) and those with ePD (b)

Full size image

Factors predictive for the development of psPD

Clinical and biological factors were analyzed to detect predictors for the development of psPD (Table 2). The p53 overexpression was found to be the most relevant predictive factor for the development of psPD (P = 0.03 by Fisher’s exact test). Seven of the eight tumors (87.5%) from patients with psPD showed overexpression of p53, as compared to 3 (30%) of the ten tumors in patients with ePD. In contrast, other factors, such as age, sex, extent of surgery, histology, surgery to RT interval, and use of FSRT boost were not significantly correlated with the early MRI response.

Table 2 Association of the development of pseudoprogression with clinical or biological factors
Full size table

Survival

Table 3 reports the effect of the first MRI finding after RT. For the entire study population, the median TTP was 8.4 months with 1- and 2-year TTP rates of 41 and 32%, respectively. The estimated median OS was 25.2 months with 1- and 2-year OS rates of 81 and 54%, respectively. The median TTP for patients with psPD, ePD, and nonPD was 7, 3.1, and 22.1 months, respectively (Fig. 2a). The TTP after ePD was inferior to that after psPD (P < 0.01) or nonPD (P < 0.01). All patients with psPD were living at last follow-up, and median OS for patients with ePD and nonPD was 10.8 and 25.6 months, respectively (Fig. 2b). The OS after ePD was worse than that for the other subgroups (P < 0.01 for each comparison). The difference in TTP and OS between patients with psPD and patients with nonPD was not statistically significant (P = 0.15 and 0.25, respectively).

Table 3 Effect of the first MRI finding
Full size table
Fig. 2
figure 2

Kaplan–Meier plots of treatment outcomes. Comparisons between subgroups divided by the early MRI response in a TTP and b OS.P is given to compare outcomes of patients with psPD and those of the patients with ePD

Full size image

Discussion

The phenomenon of “pseudoprogression” of malignant glioma was mentioned by Hoffman et al. [14] in 1979 and described by de Wit et al. [15] in 2004, and since has become well understood. It has been suggested that psPD may represent a result of treatment-related cellular hypoxia that leads to expression of hypoxia-regulated molecules from tumor and surrounding cells with a subsequent increase in vascular permeability and increased tumor enhancement [16]. In addition, vascular changes and leakiness caused by radiation, which can lead to edema and enhancement of tumor, is considered crucial for the development of psPD [17]. However, much work is needed to elucidate the underlying mechanisms of psPD. In current study, we tried to identify the potential biologic marker(s) to predict this unique phenomenon.

We observed psPD in 23% of malignant glioma patients treated with chemo-radiothearpy with TMZ. In addition, our results indicate that mutant p53 was expressed in one-third of malignant gliomas, and more importantly, significantly associated with the development of psPD. To the best of our knowledge, the present study is the first to report the predictive potential of p53 overexpression for the development of psPD. One can assume that discerning the early MRI response by such a predictor is of value in practice, because favorable survival is expected for patients with psPD, as confirmed in our study.

Contradicting result has been published about the relationship between p53 status and radiosensitivity of malignant glioma cells [18]. The clinical impact of p53 mutations in malignant glioma, in the same context, remains debatable [19–22]. With respect to a relationship between p53 and TMZ, Hermisson et al. [23] demonstrated that abrogation of p53 wild-type function strongly attenuated TMZ cytotoxicity. In contrast, an alternative pathway to induce apoptosis in p53 mutant gliomas after TMZ treatment was shown by Roos et al. [24].

Recently, studies have shown that tumor samples with methylated MGMT promoters or low active MGMT have a tendency to harbor mutant p53 [11, 25, 26], although conflicting data were reported in another study [27]. As shown by Brandes et al. [4], MGMT promoter methylation is known to positively correlate with the development of psPD; thus, our findings seem to support the relationship between MGMT promoter methylation and psPD. Taken together, these findings suggest that malignant gliomas with MGMT promoter methylation are highly sensitive to chemoradiotherapy with TMZ. Furthermore, these cells have impaired double-strand DNA break repair capability, and p53 mutation may be a surrogate marker for MGMT promoter status. However, various mutations have been found in TP53 [28], and G:C–A:T transition mutation in p53 could be related to a defect in MGMT activity, which leads to changes in G:C–A:T nucleotides during DNA replication [26].

Although patients with psPD were not followed-up long enough to calculate a median overall survival, no patient died after exhibiting psPD. The survival benefit from the addition of TMZ to radiotherapy in glioblastoma patients with methylated MGMT promoters, compared with those with unmethylated MGMT promoters, has been shown in a study by the European Organization for Research and Treatment of Cancer–National Cancer Institute of Canada [29]. Good response after chemoradiotherapy with TMZ may affect prolonged OS, and the development of psPD seems to reflect a better response of the tumor to treatment. Another possible explanation for the high OS rates in patients with psPD is the good salvage response after progression. Although the difference was insignificant, the duration of TTP in patients with psPD was shorter than in those with nonPD. Thus, it can be assumed that tumors responsive to first chemoradiotherapy will have good response to salvage treatment.

In conclusion, our findings suggest that p53 overexpression may be a potentially useful marker to predict the development of psPD in patients with malignant gliomas after chemoradiotherapy with TMZ.