PET measured hypoxia and MRI parameters in re-irradiated head and neck squamous cell carcinomas: findings of a prospective pilot study [version 1; peer review: 1 approved, 1 approved with reservations]

Background: Tumor hypoxia measured by dedicated tracers like [18 F]fluoromisonidazole (FMISO) is a well-established prognostic factor in head and neck squamous cell carcinomas (HNSCC) treated with definitive chemoradiation (CRT). However, prevalence and characteristics of positron emission tomography (PET) measured hypoxia in patients with relapse after previous irradiation is missing. Here we report imaging findings of a prospective pilot study in HNSCC patients treated with re-irradiation. Methods: In 8 patients with recurrent HNSCC, diagnosed at a median of 18 months after initial radiotherapy/CRT, [18F]fluorodeoxyglucose (FDG)-PET/CT (n=8) and FMISO-PET/MRI (n=7) or FMISO-PET/CT (n=1) were performed. Static FMISO-PET was performed after 180 min. MRI sequences in PET/MRI included diffusion-weighted imaging with apparent diffusion coefficient (ADC) values and contrast enhanced T1w imaging (StarVIBE). Lesions (primary tumor recurrence, 4; cervical lymph node, 1; both, 3) were delineated on FDG-PET and FMISO-PET data using a background-adapted threshold-based method. SUVmax and SUVmean in FDGand FMISO-PET were derived, as well as maximum tumor-to-muscle ratio (TMRmax) and hypoxic volume with Open Peer Review


Introduction
Locally advanced head and neck squamous cell carcinomas (HNSCC) that are not associated with human papillomavirus (HPV) infections have an unfavorable prognosis.Standard treatment consists of definitive chemoradiation (CRT) with radiation doses of around 70 Gray (Gy).Acute toxicity is considerable and dose limiting in this approach.Nonetheless, about 50% of patients present with local or regional recurrence after definitive CRT 1 .Re-irradiation is frequently chosen in cases with recurrence at the primary tumor site or regional lymph nodes.Due to radiation-induced long-term sequelae of the surrounding organs at risk, the re-irradiation dose is usually lower than in the primary setting.Consequently, tumors that are apparently radioresistant are treated with lower radiation doses than those applied in the primary situation.Not surprisingly, this approach is associated with a very unfavorable outcome.Especially in HPV negative tumors that are unresectable, re-irradiation is merely a palliative approach 2 .
One important factor of radioresistance is tumor hypoxia.Hypoxia is associated with a more aggressive tumor phenotype as shown by various studies of different tumor entities [3][4][5] .Additionally, when treating patients with photon radiotherapy, hypoxia leads to a decreased cytotoxic effect of irradiation due to the lower availability of oxygen radicals.This effect can be specified by calculating the oxygen enhancement ratio (OER).Usually the OER is between two and four, which further underlines the important and strong effect of hypoxia in radiotherapy 6 .One sophisticated method to measure hypoxia non-invasively is positron emission tomography (PET) with hypoxia specific radiotracers.The most commonly used hypoxia radiotracer is [ 18 F]fluoromisonidazole (FMISO) 7 .PET measured hypoxia provides a substantial and independent prognostic value in patients undergoing primary CRT for HNSCC [8][9][10] .Re-oxygenation, measured by repeated hypoxia PET during treatment, seems to be of even greater prognostic relevance in HNSCC with hardly any local tumor control in case of residual tumor hypoxia during the second week of CRT [11][12][13][14] .Given these data, we postulated that PET measured tumor hypoxia should also play an important role in local recurrent HNSCC after prior radiotherapy.To improve patient outcome by increasing re-oxygenation during CRT, a prospective pilot study was initiated to evaluate the effect of fever-range whole-body hyperthermia (FRWBH) on the tumor microenvironment in patients with re-irradiation for HNSCC.FRWBH has been shown to increase tumor perfusion in preclinical models 15,16 .The rationale of the GKH-TMM trial was to increase tumor perfusion and subsequently reduce tumor hypoxia by adding weekly FRWBH to re-irradiation.The primary endpoint of the trial was feasibility of FRWBH, which will be published separately when mature outcome data of patients are available.Secondary endpoints included changes of hypoxia and perfusion between pre-treatment and second week of fractionated CRT.Due to the lack of PET detectable hypoxia prior to treatment, we were not able to calculate these planned secondary outcome parameters.Here we report the imaging findings of the pretherapeutic PET and MRI scans within this trial.

Methods
The GKH-TMM, ARO-2018-3, study was registered at clinical trials (ClinicalTrials.govidentifier NCT03547388) and has been approved by the local Ethics committee (Charité Ethics committee, campus Virchow, EA2-047-18).All patients provided written informed consent to participate in the study and to publish results in a pseudonymized way.
The article complies with the reporting guidelines for observational studies (STROBE).

Study design
The GKH-TMM study was a prospective Phase-I study to evaluate the effect of FRWBH on the tumor microenvironment.Inclusion criteria for this study were as follows: unresectable local, regional or loco-regional recurrent non HPV-associated HNSCC with prior high-dose radiotherapy of the head and neck region (either as definitive or as adjuvant CRT or radiotherapy), time interval between previous radiotherapy and recurrence between 6 months and 5 years, complete whole-body staging without evidence of distant metastases by [ 18 F]fluorodesoxyglucose (FDG)-PET/CT, Eastern Cooperative Oncology Group (ECOG) performance status between zero and two, and age between 18 and 75 years.
The study was designed as a pilot study with ten patients, who were recruited between April 2018 and March 2020.Eligible patients were asked to take part in the trial as a complementary method to routine treatment.The evaluation of tumor microenvironment was performed by pre-therapeutic FMISO-PET in combination with magnetic resonance imaging (MRI) with diffusion weighted imaging (DWI and ADC maps) and contrast enhanced high resolution imaging (post contrast T1w StarVIBE).FMISO-PET was scheduled prior to therapy and repeated at the end of the second week of CRT in case of evidence for pre-therapeutic hypoxia.

Patients and treatment
Eight out of ten patients had pre-therapeutic FMISO-PET images.In two patients, FMISO-PET could not be performed due to logistical reasons.Out of the eight patients, seven patients underwent integrated PET/MRI, while one patient had to be examined by PET/CT due to severe claustrophobia.
All patients underwent hyperfractionated re-irradiation with two fractions of radiotherapy (1.2 Gy) per day and a minimum of eight hours between each fraction.Total treatment dose was 66 Gy, prescribed to the macroscopic tumor lesions plus 5 mm safety margin.Concomitant chemotherapy was not specified in the protocol but should preferentially include cisplatin due to its increased efficacy at mildly increased temperatures 17 .

FDG-PET/CT
Pretherapeutic FDG-PET/CT was performed with a dedicated PET/CT scanner (Philips Gemini TF 16, Philips, Amsterdam, The Netherlands).Patients were required to fast for at least 6h prior to tracer injection, and a blood glucose level ≤130 mg/dl was validated.After intravenous injection of a median of 259 MBq [ 18 F]FDG (interquartile range [IQR], 252 to 296 MBq/kg; median, 4.4 MBq/kg; IQR, 3.9 to 4.9 MBq/kg), static PET acquisition was performed after 80 min (IQR, 68 to 84 min) for 2 or 3 min per bed position from base of skull to proximal femora in supine position (matrix, 144 × 144).PET data were reconstructed iteratively using ordered subset expectation maximization (OSEM; BLOB-OS-TF) with 3 iterations and 33 subsets and time of flight (voxel size, 4.0 × 4.0 × 4.0 mm 3 ) without resolution recovery (point spread function; PSF).Random correction, scatter correction and dead time correction were also included.Attenuation correction was performed based on a non-enhanced low-dose CT (slice thickness, 5 mm).

Image evaluation
Image analysis was performed as previously published and as shortly described in the following passage using ROVER software (version 3.0.50h;ABX, Radeberg, Germany; available freely for research purposes on request) without any preprocessing of the final image data 18 .After co-registration of diagnostic FDG-PET/CT images and FMISO-PET/MRI or PET/CT images by the mutual information algorithm, correct alignment of the primary tumor volume was verified, and co-registration was corrected, if deemed necessary.The tumor (primary tumor recurrence or lymph node) was semi-automatically delineated in the PET images using a background-adapted threshold-based algorithm 19,20 .Central necrotic tumor areas or areas that were suspicious of tumor in MRI or CT were subsequently included manually to the primary tumor/lymph node volume.A reference region of interest (ROI) was delineated in the deep neck muscles contralateral to the primary tumor with a spheroid of 16 mm diameter.The ROIs of the tumor and of the muscle were transferred to the hypoxia PET for calculation of the maximum standardized uptake value (SUV max ; normalized to the body weight) of the tumor, of the tumor-to-muscle ratio (TMR) and for thresholding of hypoxic volumes, e.g. the commonly used hypoxic volume with uptake above 1.6 times of the average muscle uptake (HV 1.6 ).
The intensity of lesional contrast enhancement was rated by an experienced radiologist (9 years of experience) relative to the corresponding contralateral normal tissue using either the contrast-enhanced T1w images (StarVIBE sequence; n=7 patients) or the contrast-enhanced CT data (n=1 patient).Intensity was rated on a three point Likert-type item as "less intense", "comparable intensity" or "more intense" compared to the contralateral reference tissue.Mean ADC values of the tumor lesions were measured by the same radiologist using representative 2D region of interest placed in the tumor volume.

Statistical analysis
Statistical analysis was performed using SPSS (version 26, IBM, Armonk, NY, USA).On the basis of the small sample size, non-normal data distribution was assumed, and descriptive data were expressed as median, IQR and range, unless otherwise specified.

Results
Eight patients with available pretherapeutic FMISO-PET imaging are reported here (PET/MRI, n=7; PET/CT, n=1).Characteristics of patients and treatment are summarized in Table 1.

Hypoxia in FMISO-PET
Assessment of tumor hypoxia in static FMISO-PET images three hours post injection revealed absence of detectable hypoxia in almost all patients.The median FMISO TMR max was 1.7 (IQR, 1.3 to 1.8; range, 1.1 to 1.8).Only two patients showed hypoxic volumes that were delineable using the 1.6-fold average background muscle activity threshold (HV 1.6 ).The median HV 1.6 of all 8 patients was 0.05 ml (IQR, 0 to 0.33 ml; range, 0 to 7.3 ml).No lesion showed a TMR max >2.0.Notably, especially the largest recurrent lesions did not show any detectable hypoxia (HV 1.6 = 0 ml).
Details of FDG-PET and FMISO-PET parameters for all patients are shown in Table 2. Figure 1 shows exemplary FDG-PET/CT scans and FMISO-PET/MRI scans.

Contrast enhancement and ADC values
Tumor contrast enhancement was less intense compared to the contralateral reference in four of eight patients (lymph node only, n=1; primary tumor recurrence ± lymph node, n=3; Table 2).In the remaining four patients, enhancement was more intense, at least in parts of the lesion, compared to the reference tissue (primary tumor ± lymph node, n=4).

Discussion
Here we report pretrial data from the first trial that evaluated tumor hypoxia by specific hypoxia PET in previously irradiated, recurrent HNSCC.Data on hypoxia measured by PET tracers in the recurrent disease situation after radiotherapy is absent.To the best of our knowledge, there is only one study that investigated various tumor locations and histologies with hypoxia PET.In the mentioned study, the authors describe one patient with recurrent adenocarcinoma of the uterus who showed a relatively high uptake of FMISO.It is not clear if this patient was previously treated with radiotherapy/CRT or presented recurrence after initial surgery 21,22 .
Very surprisingly, the incidence of hypoxia, determined by FMISO-PET, was extremely low in our study investigating recurrent HNSCC.The largest analysis of hypoxia PET images from five European centers established the hypoxia parameter TMR max 2.0 as a suitable value to distinguish high-and low-risk treatment-naive patients.The prevalence of hypoxia (defined as TMR max > 2.0) was relatively high in primary tumors.PET hypoxia was found in 48% of all patients; among those who received FMISO-PET, 61% showed hypoxic tumors/lesions 18 .In the present trial of previously irradiated HNSCC, none of the eight included patients showed hypoxic tumors according to these cut-off values, despite the fact that patients with relatively large recurrent lesions were included (see Table 2).This finding may indicate that microenvironmental mechanisms of tumor radioresistance might be very different between treatment-naïve and previously irradiated tumors.This also seems to be the case with genetic alterations 23,24 .Dose painting for re-irradiation does not seem feasible in FMISO-PET due to the low tracer uptake, i.e. low rates of hypoxia in this patient collective [25,26,p.33].In each patient, FDG uptake of the recurrent lesion is extensive and high, while specific FMISO uptake was absent (HV 1.6 ≤1.0 ml).
In addition to late static FMISO-PET images, dynamic acquisition from 0 to 40 min post injection has been proposed.Combining dynamic and static PET images for a voxel-wise 2-compartment model of [ 18 F]FMISO accumulation, Thorwarth et al. were able to identify patients with treatment-naïve HNSCC with either favorable or very poor local tumor control after radiotherapy.The derived dynamic parameter M FMISO was superior to TMR max in predicting local tumor control 27 .Dynamic FMISO-PET acquisition was not available in our patient collective.However, it is questionable if dynamic data would provide any added value in lesions that are negative on late FMISO images.
In addition to mostly absent hypoxia, ADC values were relatively high in all tumors with a median of >1,000 × 10 6 mm 2 /s, which corresponds to only mildly restricted diffusion.
Only in one tumor (patient #7), the average ADC was below 1,000 × 10 6 mm 2 /s.These MRI findings are consistent with the literature [28][29][30]  Moreover, ADC values as high as 1,000 × 10 6 mm 2 /s were only observed in tumors with hypoxic volumes <10% of the total tumor volume 32 .In contrast, a positive correlation was reported in murine melanoma tumors (but imaging was performed ex vivo) 33 .Swartz et al. and Carmona-Bozo et al. found no correlation between hypoxia and ADC values in oropharyngeal and breast cancer lesions, respectively 34,35 .
Contrast enhancement (StarVIBE MRI sequence or contrastenhanced CT) was atypically low in 4/8 lesions, and hypoxia was also absent in these tumors (Table 2).Among the 4/8 lesions with substantial contrast enhancement (at least in the tumor periphery), some showed small hypoxic proportions as determined by the FMISO HV 1.6 .Gerstner et al. found that high lesional cerebral blood flow in glioblastoma quantified by dynamic susceptibility contrast MRI correlated positively with the hypoxic volume in FMISO-PET.The authors postulated that abnormal tumor vasculature contributes to hypoxia 36 .This may explain the lack of substantial and extensive contrast enhancement in the current mostly non-hypoxic tumors.However, the current sample size was too small to conclude if there is a negative correlation between contrast enhancement and tumor hypoxia.Dynamic contrast MRI sequences were not evaluated in this study, and contrast kinetics -as described by 36 -could not be assessed.
In summary, the current pilot study implies that relevant hypoxia, which is detectable by static FMISO-PET, may not be as prevalent among recurrent lesions of HNSCC as expected.However, studies with substantially larger sample sizes would be required for a definite statement or more differentiated analyses (e.g., dependence of hypoxia on the type and localization of the recurrent lesion).This is also true for the investigation of a possible association of low tumor hypoxia with high ADC values and low contrast enhancement in these recurrent lesions.
The present manuscript deals with a very interesting issue reporting preliminary data on the use of a PET radiotracer, 18F-FMISO, to measure hypoxia in head and neck squamous cell carcinoma patients with relapse after previous irradiation.Eight patients have been included, 7/8 studied by FMISO PET/MRI and 1/8 by using PET/CT.
As for strengths, the paper is very well written and methodologically correct.
In addition, although 18F-FMISO is a well-established prognostic factor in head and neck squamous cell carcinomas treated with definitive chemoradiation, data on prevalence and characteristics of PET measured hypoxia in patients with relapse after previous irradiation are still missing.This paper report imaging findings of pretherapeutic PET and MRI scans within a prospective trial.
As weakness, the number of patients included is very limited and the population is heterogeneous.In addition, two different scanners have been used (7 PET/MRI and 1 PET/CT).The rationale and the use of FDG in this preliminary evaluation has not been definitely defined and commented.
It should be also clarified why

Jacqueline Kelly
Department of Therapeutic Radiology, Yale School of Medicine, New Haven, CT, USA In this manuscript, the authors explore the novel and important question of hypoxia levels in recurrent HNSCC that has previously been irradiated with high-dose RT.Because hypoxia has been shown to be prognostic in the definitive HNSCC RT setting, and as re-RT is toxic and may be effective in only very select patient populations, the knowledge gained from this study could be used to select appropriate candidates who may be more likely to have a durable response to re-RT.Surprisingly, the incidence of hypoxia was extremely low in the present study, albeit in a very limited number of patients in a heterogeneous patient population.
The paper is well-written, with clear methodology, presentation of the results, and an appropriate discussion.In viewing Table 1, it appears initial treatment was uniform, with def CRT patients receiving 70.4 Gy and adj CRT receiving 63.9.Since these doses are clearly not achieved using the commonly used 2Gy/fraction, it may be worth mentioning the fractionation schedule used with the initial treatment course.Altered fractionation schedules have at least theoretical advantages relative to OER effects and thus this should simply be noted.
I would also recommend in the conclusion reiterating that these findings apply to only the HPV(-) HNSCC population.
Overall, a soundly conducted study, unfortunately with conclusions limited by a very low number of patients.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound?Yes The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com

Table 2 . FDG-PET, FMISO-PET and MRI parameters
. Abbreviations: MTV = metabolic tumor volume, SUV max = maximum standardized uptake value, SUV mean = mean standardized uptake value, TMR max = maximum tumor-to-muscle ratio, HV1.6= hypoxic volume with a threshold of 1.6 times the average background/ muscle uptake; ADC = apparent diffusion coefficient.
*Only a single voxel above the threshold . Hwang et al. reported average ADC values of 1,200 × 10 6 mm 2 /s in patients with recurrent HNSCC after initial treatment compared to an average of 1,650 × 10 6 mm 2 /s in lesions corresponding to post-therapeutic changes 30 .Previous reports on the association between tissue hypoxia in different tumor entities and properties in DWI or corresponding ADC maps are inconsistent.Hino-Shishikura et al. reported a negative correlation between ADC values and the SUV max and tumor-to-background ratios in brain tumors obtained with hypoxia-specific PET imaging 31 .Hompland et al. demonstrated a weak negative correlation between ADC values and hypoxic fraction assessed by pimonidazole tissue staining in malignant melanoma xenografts.

Is the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? Yes If applicable, is the statistical analysis and its interpretation appropriate? Yes Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes Competing Interests:
1.6-fold muscle has been considered to define hypoxic volume.In literature other values have been also reported (i.e.1.2; 1.4)No competing interests were disclosed.

have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above. hypoxia
volumes inside and outside the macroscopic tumor lesions.For better ilustration we included images of the two patients with (small) HV1.6 volumes inside the macrospopic tumor (#4 and #6) as supplementary information and hope that this visualizes that a true hypoxic volume can probably only defined in patient #6.This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.