Untreated HNSCC patients with nodal metastases underwent pretreatment MMI with 1
H-MRS, DCE-MRI and 18
F-FDG PET. Although there is no direct biological one-to-one link between these non-invasive imaging techniques, we found strong correlations between selected imaging parameters (Cho/W, 18
F-FDG SUV parameters). Two recent studies showed that the combination of DCE-MRI, 18
F-FDG PET, and 18
F-fluoromisonidazole (FMISO) PET has potential for assisting treatment planning for HNSCC patients. However, both studies were focused on hypoxia, and they did not correlate the data obtained from DCE-MRI and 18
F-FDG PET (12
). The results of our study indicate that data from 1
H-MRS, DCE-MRI and 18
F-FDG PET in patients with HNSCC are complementary, and not competitive.
A strong positive correlation between Cho/W and 18
F-FDG TLG was observed. High Cho/W concentrations are indicative of an increased membrane choline phospholipid metabolism suggesting high proliferation (25
). Khan et al. recently studied patients with HNSCC with both 18
F-FDG PET (glucose metabolism) and 11
C-labeled choline PET (phospholipid metabolism) and observed a significant correlation between 18
F-FDG and 11
C-choline SUVs (26
). Our results support their inference that increased glucose metabolism is related to increased cellular proliferation. As neither SUVmax nor SUVmean displayed a significant correlation with Cho/W, tumor volume (required for calculating TLG) might be important.
A strong negative correlation was observed between Cho/W and std(kep
) as well as between Cho/W and std(ve
). Standard deviation measures describing the width of the pixel histogram distribution are indicative of the tumor heterogeneity (22
). Our results suggest that areas of dense tumor cell population typically have high choline concentrations, due to a high membrane turnover, whereas heterogeneous head and neck tumors contain areas of low proliferation and often highly necrotic regions, as was suggested by our study. It has previously been shown in patients with gliomas that necrotic tissue has significantly lower Cho concentrations than non-necrotic tissue (27
). Other processes such as hypoxia might also play a role in heterogeneity (15
), but hypoxia measurements were not obtained in this study. In a recent study, we employed FMISO PET/CT and DCE-MRI in patients with HNSCC, and showed that hypoxic nodes are poorly perfused (i.e., have significantly lower Ktrans
values) compared with nonhypoxic nodes (15
In accordance with the literature (28
), MRI tumor volume, correlated strongly with SUVmax, SUVmean, and TLG from 18
Pretreatment DCE MRI parameter std(Ktrans
) and 18
F-FDG measure SUVmean were significant predictors for short-term response as assessed by radiological evaluation 3–4 months after completion of treatment. This is in accordance with previous studies, in which pretreatment DCE-MRI (29
) and 18
F-FDG PET (30
) were predictive of response to therapy in HNSCC. Pretreatment DCE-MRI and 18
F-FDG PET might therefore help physicians with evidence-based treatment planning. As the separate predictive performance of both significant predictors (std(Ktrans
) and SUVmean) was inferior to the performance of both predictors combined, a synergic action of the two modalities was demonstrated.
Our study has some limitations. Firstly, the number of patients was low (n=16). Secondly, the minimum ROI that can be achieved by 1
H-MRS is dependent on the signal-to-noise ratios of the metabolites studied. We studied choline only, as creatine was not visible. Studying lactate would require another specialized sequence for data acquisition (31
), which was beyond the scope of this clinical study. The use of surface coils instead of phased-array coils would have been ideal for acquiring data from superficial tumors; However, in the present study, 1
H MRS was part of a clinical exam, and thus switching coils during the examination was not feasible. Though even with optimal techniques, 1
H MRS of necrotic tumors will still be a challenge, as such tumors may show very low choline concentrations, resulting in high CRLB values.
Finally, another limitation was that a pixel by pixel analysis of DCE-MRI and 18F-FDG PET images was not performed. It should be noted that the patient orientations during the two imaging (MRI and PET) examinations were different. DCE-MRI was performed as part of a diagnostic clinical MRI examination whereas PET was in most cases performed as part of a radiation treatment planning study. To enable a pixel-by-pixel comparison, the patient positioning would need to be exactly the same for MRI and 18F-FDG PET.
The present feasibility study shows interesting a priori results with DCE-MRI, 1HMRS and FDG-PET data. Such pretreatment data may have translational applications in three areas: treatment planning, prediction of short-term treatment response or outcome, and monitoring treatment. MMI data shows in-vivo the heterogeneity of the nodal metastases in HNSCC patients and our initial results have shown that it can predict short-term response. In the future, if pretreatment MMI data can help distinguish tumors with a “good” from those with a “poor” prognosis, its use may allow patient-specific treatment. Specifically, it will help identify patients at risk earlier so that they can be considered for treatment with anti-angiogenic agents, hypoxia-targeting therapy or gene therapy. Additionally, MMI data will further improve our understanding of tumor biology in-vivo and help us unravel new treatment strategies.