Introduction

Oxygen deficiency, termed hypoxia, is a hallmark of solid tumors that encourages tumor angiogenesis, genetic instability and metastasis ¹. It is an adverse prognostic factor in cervical cancer ². MRI is sensitive to changes in the spin-lattice relaxation rate (ΔR₁) following delivery of 100% oxygen ³ delivery in tissues ⁴. Previous pre-clinical evaluation validated a combined OE-MRI and dynamic contrast enhanced (DCE)-MRI biomarker (the volume of tissue that is perfused and also refractory to oxygen gas challenge; termed pOxy-R) as a measurement of tumour hypoxia ⁵. Subsequent translation in patients with non-small cell lung cancer showed that pOxy-R identified and mapped hypoxia in human tumours, and detected hypoxia modification following chemoradiotherapy ⁶. We sought to extend this technology to patients with uterine cervix carcinoma and report a) quality assurance (QA) steps that were optimized in healthy volunteers, and b) initial data evaluating pOxy-R as a biomarker of hypoxia in patients with cervical cancer.

Methods

Overview: Research ethics approval was obtained (REC: 20/NW/0377). Patients provided written informed consent. Data was obtained for 5 healthy volunteers and 3 patients with stage IIB-IVA squamous cell carcinoma of the cervix. Treatment involved weekly cisplatin chemotherapy prescribed at 40 mg/m2. Combined chemoradiation and brachytherapy were prescribed to reach a final dose of 85 to 90 Gy EQD2.

Acquisition: Imaging was acquired in the sagittal plane on a 1.5T Philips Ingenia MR-RT (Amsterdam, Netherlands). All imaging was in the sagittal plane. The reconstructed matrix was 128 x 128 pixels and the resolution was 3 x 3 x 6 mm3. Gases were delivered using a tight-sealed, non-rebreathing Intersurgical EcoLite™ adult facemask (Intersurgical Ltd, UK). Switching between the gases was controlled via the Low Flow Air-Oxygen Blender (Inspiration Healthcare, UK). (1) OE-MRI: 3D Inversion Recovery Turbo Field Echo (IR-TFE) sequence to map T₁ (TR = 2.2 ms; TE = 0.66ms; TI = 100, 500, 1100, 2000, 4300 ms; ⍺ = 4°). Switched to dynamic sequence with medical air breathing (21% O2) followed by oxygen breathing (100% O2) at the 26th time point (TI = 1100ms; temporal resolution = 12s). (2) DCE-MRI (patients only): 3D mDIXON fast field echo (FFE) sequence to map T₁ (TR = 3.2 ms; TE₁ = 1.7 ms; TE₂ = 2.2 ms; ⍺ = 2°, 8°, 12°). Switched to dynamic sequence with a bolus of 0.1 mmol/kg gadoterate meglumine (Dotarem®, Guerbet, France) injected at 3 ml/s at the 8th time point (⍺ = 8°; temporal resolution = 3.8s).

Analysis: For OE-MRI voxel-wise ΔR₁ was calculated ⁷, where: ΔR₁ = R₁ while breathing oxygen (mean of last 10 time points on 100% oxygen) minus R₁ on breathing air at the start (mean of last 10 time points on medical air). Repeatability was assessed in healthy volunteers by measuring the within-subject coefficient of variation (wCV). In patients, the same OE-MRI analysis was performed. In addition, the biomarker pOxy-R was calculated, by firstly identifying perfused voxels on DCE-MRI and then classifying voxels as hypoxic (pOxy-R) when DCE-MRI integrated are under the gadolinium enhancement curve at 60 seconds was >0 (one sided paired sample t-test, p<0.05) and ΔR₁ was refractory to change. Analyses were performed in MATLAB R2020a (Mathworks).

Results

The entire MRI protocol was well tolerated in all healthy volunteers and patients, with no adverse events. ΔR₁ plots in healthy volunteer tissues (uterine body and right psoas muscle) are shown in Figure 1, along with double baseline repeatability. Mean ΔR₁ Uterine Body = 0.032 ± 0.014 s-1 (p<0.00005). In distinction, mean ΔR₁ Muscle = 0.0022 ± 0.0072 s-1 (p=0.35). This determined that the uterine body would be used as an indicator of successful oxygen delivery in a QA step.

Patient data showed mean ΔR₁ Uterine Body values in patients were 0.021 ± 0.012 s-1 (baseline), 0.035 ± 0.025 s-1 (week 3) and 0.039 ± 0.026 s-1 (week 5) (Figure 2) indicating technical success in all scans. Figure 3 shows illustrative anatomical T2w images paired with combined OE-MRI and DCE-MRI maps in one patient at baseline and weeks 3 and 5, revealing spatial heterogeneity in hypoxia. The tumour pOxy-R reduced in patients 1 and 3 at week 3 and remained reduced at week 5, whereas patient 2 had minimal changes during treatment. Both individual and cohort changes are summarized (Figure 4).

Discussion and conclusion

OE-MRI is a promising technique for hypoxia evaluation, but has low SNR and so QA is useful to exclude technical failures. Healthy volunteer imaging supported using the uterine body as an independent QA reference region and this was then applied to patient data. Muscle represents a poor choice given the low mean ΔR₁ and high within-subject coefficient of variation. OE-MRI and DCE-MRI were well tolerated in three patients with cervical carcinoma who had serial mapping of hypoxia while undergoing chemoradiotherapy. Initial results suggest that two patients had therapy-induced reduction in hypoxia within 3 weeks that persisted to week 5. Further recruitment is ongoing, but initial data are encouraging to support OE-MRI as a useful non-invasive technique to quantify and map hypoxia modification in patients with cervical cancer.

Acknowledgements

No acknowledgement found.