Prediction of Medulloblastoma survival using the Water Proton Resonance Frequency
Ben Babourina-Brooks1, Sarah Kohe1, Andrew Peet1, and Nigel Davies2

1University of Birmingham, Birmingham, United Kingdom, 2University Hospitals Birmingham, Birmingham, United Kingdom

Synopsis

The rate of improvement in survival, among children with brain tumours, has decreased in recent years and novel prognostic markers that may contribute to treatment stratification and improved outcomes are required. Medulloblastoma tumours have a high rate of mortality due to their aggressive nature, however some do not. MRS can measure absolute water proton resonance frequency (PRF), which is sensitive to temperature and other features of the tumour microenvironment. This could be utilized as a potential prognosis marker, however has not been explored. This study assessed the water PRF as a predictor of survival in a medulloblastoma patients, using MRS at diagnosis.

Introduction

The rate of improvement in survival, among children with brain tumours, has decreased in recent years and novel prognostic markers that may contribute to treatment stratification and improved outcomes are required. Medulloblastomas have the highest rate of mortality due to their aggressive nature, WHO grade IV, compared to other common paediatric tumour types. However not all medulloblastomas have a high risk of mortality and establishing the non-invasive markers of prognosis in this group of tumours is of particular interest1. In addition to providing characteristic metabolite levels2, MRS can measure absolute water proton resonance frequency (PRF). Water PRF is sensitive to temperature and other features of the tumour microenvironment and has been shown to be useful for characterisation of common childhood brain tumour types3. However, the potential for variations in water PRF within tumour types to non-invasively predict prognosis has not been explored. The aim of this retrospective study was to investigate water PRF as a predictor of survival in a cohort of medulloblastoma patients using MRS at diagnosis.

Methods

Single-voxel MRS data were acquired using a 1.5T Siemens system (PRESS, TR 1500ms, TE 30ms) in 45 histology confirmed medulloblastoma brain tumour patients. The accrual period was between September 2003 and December 2012 and patients were followed up until August 2015. In this time span a number of treatment protocols were used. The general principles were that a maximal surgical resection of the primary tumour was followed by adjutant treatment: chemotherapy and craniospinal radiotherapy for older children, focal radiotherapy for younger ones. Spectra were analysed using jMRUI and the AMARES tool4. The weakly suppressed water (H2O, 4.68ppm) and total choline (tCho, 3.22ppm) peaks were fitted with Lorentzian lineshapes and the water PRF calculated relative to tCho (δH2O(tCho)). The Cho reference was chosen since it was prominent in all tumour spectra and not highly affected by macromolecules. The median δH2O(tCho) value was used as the cut-off for a Kaplan-Maier survival analysis. A log-rank test was used to investigate the significance of the difference in survival between the groups. Histopathological subtypes and clinical features related to survival, including Chang M staging, sex and age, were compared between the high and low δH2O(tCho) groups and regression analysis was performed to investigate correlations.p<0.05 was used as the threshold for significance in all tests.

Results and Discussion

A total of 40 patients with medulloblastoma were eligible for the study and 18 had died by the end of the study period. Kaplan-Maier analysis (Figure 3) shows good separation of the high and low δH2O(tCho) groups throughout the five year time span with significantly poorer survival in the low δH2O(tCho) group (chi square 8.4, p <0.01). Based on previous studies this could indicate that more aggressive medulloblastomas have a higher temperature or lower protein concentration or a contribution from both3,5. Table 1 shows a comparison of patient characteristics that may affect survival between high and low δH2O(tCho) groups. There was no significant difference in median age between the groups and a subsequent analysis showed there was no survival difference based on age in this cohort.Sex of the patient was also not found to be a significant survival factor in this study, p= 0.625.The Chang M staging values did not correlate with the water PRF shift and the mean M stage value was not significantly different between the deceased and alive patient groups in this cohort. Histological subtypes of medulloblastoma are known to provide indicators of survival6, specifically the desmoplastic nodular and large cell anaplastic tumour sub-types, indicating good and poor prognosis, respectively. Seven out of the eight desmoplastic nodular tumours survived and all were in the high water PRF group. All three of the large cell tumour patients died, and two of these had low water PRF potentially predicting poor prognosis. The third patient was close to the cut off value, 4.694ppm. All reported myc status’ for the cohort were negative in this study.The possible effect of differences in treatment regime on survival in this cohort and any other potential bias on the survival analysis based on water PRF will be investigated.

Conclusion

This study suggests that absolute water PRF may be a useful non-invasive marker of prognosis in medulloblastoma. This novel biomarker, related to the tumour microenvironment, can be measured using common MRS methods independently of and complementary to metabolite profile measurements. A More detailed investigation of the added-value of the water PRF measure compared with known prognostic factors and its potential to improve treatment stratification are warranted.

Acknowledgements

No acknowledgement found.

References

1. Wilson M et al. Clin Cancer Res; 2014:20(17), 4532-4539.

2 Davies NP et al. NMR Biomed. 2008; 21: 908–918.

3.Babourina-Brooks B et al. NMR Biomed. 2014:27(10):1222-9.

4. Barkhuijsen, H et al. J MagnReson 1985;61: 465-481.

5 Babourina-Brooks B et al. NMR Biomed. 2015:28(7) 792-800.

6. Eberhart CG et al . CANCER.2002: 94(2) 652-560.

Figures

Figure 1: The voxel placement on a T1 weighted image of a SVS sequence for a paediatric tumour (left) and the mean spectrum of the medulloblastoma cohort (right).

Figure 2: Kaplan-Maier analysis for δH2O(tCho) based on a median value cut-off.

Table 1: A summary of patient characteristics for groups based on the δH2O(tCho) median cut offs. Percentage of males, Chang M0 tumour staging, Histopathological tumour types and the median age in each group was assessed.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2430