*3.3. Correlation between 68Ga-Nitroimidazole PET/CT, 18F-FDG PET/CT Imaging and Immunohistochemistry*

There was no statistically significant correlation between the hemoglobin levels, 18F-FDG SUVmax, 68Ga-Nitroimidazole imaging and immunohistochemical analysis. The median SUVmax and TMR on both 18F-FDG and 68Ga-Nitromimidazole at both time points did not demonstrate any statistical difference in hypoxic and non-hypoxic states as assessed by immunohistochemistry. When considering the hypoxia score, we found that five of the seven patients considered positive for hypoxia on immunohistochemistry, displayed uptake above background on the 68Ga-Nitroimidazole scan. Table 3. Demonstrates the results of the Kruskal–Wallis test assessing the difference between quantitative parameters

on both 18F-FDG and 68Ga-Nitroimidazole PET/CT imaging in hypoxic and non-hypoxic states. There was no significant association between the presence of metastasis on 18F-FDG and the presence of hypoxia both on imaging and immunohistochemistry. None of the variables under study proved significantly different between those patients presenting with distant metastasis versus those who without (*p* ≥ 0.162). When assessing the HTV, we found no significant difference in the HTV of patients with positive results on immunohistochemical staining and those with negative results. Importantly we found a strong negative correlation between HTV and immunohistochemical staining (r = −0.660; *p* = 0.019). Additionally, hypoxia positivity proved moderately and significantly correlated to SUVmean values of the primary tumor derived from the early and delayed 68Ga-nitroimidazole PET examinations r = 0.531 (*p* = 0.42) and r = 0.580 (*p* = 0.024), see Figure 4. Additionally, upon dichotomization of the primary tumor hypoxia score (group 0, scores: 0, 1 and 2 and group 1, scores: 3, 4, 5 and 6) TMR derived from the early 68Ga-nitroimidazole PET examination proved significantly higher in tumors with score 1 versus those with score 0 (13.2 ± 7.3 versus 8 ± 13.2, respectively).

**Figure 3.** The immunohistochemical staining of HIF-1α expression in the endocervical biopsy specimen of the patient in Figure 2. Well differentiated squamous cell carcinoma ×20 magnification. The tumor areas with well differentiated cells (*yellow arrow*) with brown staining, demonstrate HIF-1α expression, while the areas with other cell types (*orange arrow)* demonstrate little to no HIF-1α expression.

**Table 3.** SUVmax and TMR from the 18F-FDG and 68Ga-Nitroimidazole PET/CT scans correlated with the HIF-1α expression.


K: Kruskal–Wallis test, 68Ga-Nitroimidazole maximum standardized uptake value (68Ga-Ni SUVmax): 68Ga-Ni SUVmax (1) and 68Ga-Ni SUVmax (2): SUVmax at 30 and 60 min post tracer injection, respectively; 68Ga-Nitroimidazole tumor to muscle ratio (68Ga-Ni TMR): 68Ga-Ni TMR and 68Ga-Ni TMR 2 are the tumor to muscle ratio at 30 and 60 min post tracer injection, respectively.

**Figure 4.** Scatterplot showing the relationship between the hypoxia score (data shown in the X-axis) and SUVmean of the primary tumor derived from the delayed 68Ga-Nitroimidazole PET images.

#### **4. Discussion**

The presence of hypoxia in solid tumors has a bearing on treatment and outcomes; therefore, non-invasive modalities that can map it are essential for therapy planning. Several PET tracers have been utilized for this purpose in various cancer entities with variable results. Our study aimed to image hypoxia in cervical cancer lesions/tumors with 68Ga-Nitroimidazole which is more hydrophilic and has demonstrated improved properties compared to 1st, 2nd and 3rd generation 18F-labelled nitroimidazole tracers. The biodistribution was as expected with intense activity seen in the kidney and urinary bladder because of the hydrophilicity. The qualitative assessment revealed that two-thirds of the patients had uptake in the primary tumor above background. This is similar to the >50% rate of hypoxic tumors in studies using invasive needle electrode measurements [6,7,28]. In a study of 38 women, Dehdashti and colleagues found discernable 60Cu-ATSM uptake in all but one patient [29]. Another study using 18F-FMISO also detected focal areas of uptake in all 16 patients [30]. Although the patient population in both studies was similar to ours (locally advanced tumors), the differences seen in the numbers of patients with uptake may be related to the differences in the tracers and tracer kinetics as well as interpretation criteria. Most studies on hypoxia imaging (pre-clinical and clinical) report on the SUVmean or none at all and we found the SUVmean on the delayed 68Ga-Nitroimidazole to be strongly correlated to the positivity on immunohistochemical staining. This finding is most likely related to the heterogeneity of hypoxic regions within a tumor and may explain why most papers on this subject have reported on the SUVmean as opposed to the SUVmax. The median SUVmax in our series was 3.63 and 3.27 at 30 and 60 min, respectively, which is comparable to that reported in the clinical work using 18F-FMISO in cervical cancer patients [30]. The other parameter that has been more commonly reported on, in the context of hypoxia imaging using different tracers is the TMR or TBR. Preclinical studies using 68Ga-labelled nitromidazoles have demonstrated that TMR and TBR were highest at 1 h post imaging. We confirmed this finding as we found a median TMR of 8 ± 4.8 and <sup>10</sup> ± 5.1 at 30- and 60-min post injection, respectively. Initial pre-clinical work on 68Ga-Nitroimidazole revealed median TMRs as high as 7.41 ±1.12, 5.70 ± 2.5, 5.64 ± 0.8 [19–21]. The work on 64Cu-ATSM also revealed TMR 7.3 ± 1.8 [31]. These high TMRs contrast with the TMRs when using 18F-labelled tracers and this may be attributed physical properties of these tracers which result in faster clearance from background tissues. We found no significant difference in the TMR and TBR between the 30 min and 60 min time points, suggesting that images may be obtained as early as 30 min post tracer injection without significantly compromising the image quality. Interestingly we also found higher TMRs on the early 68Ga-Nitroimidazole scan in patients with hypoxia on immunohistochemistry

than those with no hypoxia. This may further support early imaging. This is in contrast to 18F-FMISO which has prolonged imaging times up to 2 h post tracer injection.

There is an abundance of literature demonstrating hypoxia in cervical cancer lesions using 18F-labelled tracers and 60/64Cu-ATSM, however very few of them correlated their findings to immunohistochemistry which may be considered the gold standard. There are various genes that are upregulated in hypoxic environments including vascular endothelial growth factor (VEGF), Carbonic anhydrase IX (CAIX) and osteopontin to name a few. In a study of 44 women with advanced cervical cancer, the authors found no correlation between the expression of HIF-1α and tumor oxygenation as detected by an Eppendorf device [32]. Vercellino and colleagues assessed the feasibility of imaging hypoxic lesions with 18F-FETNIM and correlated their findings to osteopontin which is also upregulated in hypoxic environments. They found no correlation between levels of osteopontin and 18F-FETNIM uptake. Similarly, we also failed to find a correlation between the HIF-1α expression on immunohistochemistry and 68Ga-Nitroimidazole imaging. We found that almost half of the patients with any level of HIF-1α expression on immunohistochemistry had discernable hypoxia on PET imaging, while five of the seven patients with positive hypoxia scores had uptake on 68Ga-Nitroimidazole PET. This is contrary to the study by Grigsby et al. who correlated VEGF, CAIX, cyclo-oxygenase-2 (COX-2), epidermal growth factor and apoptotic index with 60Cu-ATSM PET imaging of tumor hypoxia. Immunohistochemical markers were expressed in most, if not all the tumors seen to be hypoxic on 60Cu-ATSM PET imaging [33]. Immunohistochemical analyses are fraught with challenges including operator dependence and sampling issues etc., and this may be a reason for the lack of correlation of these findings with imaging as seen in our study.

We believe there are unique patient groups or outcomes that deserve a special mention because of their interesting or unexpected findings. Two patients, namely patient 4 and 5 (Table 2) demonstrated hypoxia on immunohistochemistry, however their 68Ga-Nitroimidazole PET scans demonstrated no discernable areas of uptake. The reasons for the above may be related to size of the tumor which may render the lesion/s prone to partial volume effect or being missed because of the inherent resolution limits of the imaging unit. In three patients (7, 9 and 13), we note the reverse, wherein the despite lack of HIF-1α, the 68Ga-Nitroimidazole PET scan demonstrated uptake above background tissues. This outcome may be related to sampling errors for the HIF-1α staining in view of the heterogeneous nature of hypoxia.

Most oncology guidelines recommend the use of 18F-FDG PET/CT in the staging and restaging of most malignancies including cancer of the cervix [34]. Often, 18F-FDG PET/CT images are used for the visual estimation of uptake of hypoxia tracers within tumor lesions. Furthermore, in-vitro studies have demonstrated that hypoxia results in increased FDG uptake [35–37]. There are several studies which have correlated the findings on hypoxia PET imaging with those from metabolic imaging. Grigsby et al. found no correlation between uptake parameters on 18F-FDG and 60Cu-ATSM PET scans, however they did note a significant correlation between the presence of 18F-FDG positive lymph nodes and the findings on 60Cu-ATSM PET [33]. The lack of correlation between the metabolic and hypoxia imaging has been also demonstrated in other tumor entities [38,39]. Similarly, we could not demonstrate any correlation between PET-derived parameters (SUVmax, SUVmean, TMR and TBR) on the metabolic and hypoxic imaging. We also found no correlation between the presence of any metastasis on 18F-FDG PET and 68Ga-Nitroimidazole. The hypoxic volume was always less than the metabolic tumor volume in our patient cohort. This is in support of the reports of heterogeneity of hypoxia, therefore further highlighting the importance of hypoxia mapping with imaging studies.

We found a significant strong negative correlation between HTV and immunohistochemistry (r = −0.660; *p* = 0.019). This finding was not anticipated. We postulate that this finding may be due to several factors including tumoral environmental issues and technical factors. We believe that other factors besides upregulation of HIF-1α are at play in the hypoxic environment. A possible theory is the presence of a feedback mechanism

between markers/genes within the tumor and HIF-1α. Another plausible explanation may be related to the different isoforms of HIF-α, as it has been shown that there are at least three, namely HIF-1α, HIF-2α and HIF-3α [40]. These α-subunits are regulated at the protein level by oxygen dependent mechanisms. It has been shown that HIF-1α expression is higher in acute hypoxia whereas that of HIF-2α is higher in chronic hypoxia [41,42]. This finding calls for larger studies with modified and improved protocols to be conducted as to unravel the association we found.

To the best of our knowledge, this is the first study to assess this novel 68Ga-Nitroimidazole tracer in patients with cervical cancer. This very fact may prove to be a limitation since there are no prior studies to compare our findings to, therefore most of our comparisons are with pre-clinical studies, other 18F-labelled or the 64Cu-labelled hypoxia tracers. We could not draw any strong conclusion because of the small sample size. The main excretory pathway of the more hydrophilic 68Ga-Nitroimidazole is through the renal system. This resulted in intense bladder uptake that may pose as a challenge for pelvic malignancies. Despite urinary catheterization of the majority of our patients, the urinary bladder was often incompletely drained, and this may pose a challenge in the interpretation of the scan. Therefore, for pelvic malignancies, a tracer with minimal renal excretion or the employment of radiomics, may optimize the interpretation of the findings. The immunohistochemistry was performed on biopsy specimen that were collected at initial diagnosis and this may misrepresent the true status of hypoxia in the entire tumor lesion. Lastly, immunohistochemical analysis is highly operator dependent and this too may pose as a limitation in our study. Future studies with pre-operative patients that will undergo surgery and subsequent immunohistochemistry on more representative samples of the surgical specimen may yield improved results. While it was our desire to perform survival analysis, local factors e.g., lack of centralized health informatic systems, proved to be a challenge. This information may have added further insight in this area.

#### **5. Conclusions**

We found that there was no difference in the tumor uptake on 68Ga-Nitroimidazole PET/CT between early (30 min) and delayed (60 min) imaging, which may suggest that imaging can be acquired early without compromising the tumor to background ratio. Although we detected hypoxia in two-thirds of the patients on 68Ga-Nitroimidazole PET imaging, we found no significant relationship between HIF-1α and clinicopathological features or 18F-FDG and 68Ga-Nitrimidazole PET/CT parameters. However, we found higher TMRs and SUVmeans in patients with hypoxia as assessed on immunohistochemistry. Furthermore, there was a negative correlation between HTV as assessed by PET hypoxia imaging and HIF-1α staining. Further, larger studies are required to determine the prognostic value of using 68Ga-Nitroimidazole PET imaging to predict the pathological and prognostic course of cervical cancer.

**Author Contributions:** Conceptualization, K.M.G.M. and M.M.S.; methodology, M.M.S., M.B., J.M.J., J.M., C.D., M.V. and K.M.G.M.; formal analysis, K.M.G.M., C.v.d.W., A.M. and G.P.; investigation, K.M.G.M., I.O.L., L.C.M., K.N.H., H.N. and J.R.; resources, M.M.S. and J.M.J.; data curation, K.M.G.M.; writing—original draft preparation, K.M.G.M.; writing—review and editing, K.M.G.M., I.O.L., M.M.S. and M.V.; visualization, K.M.G.M. and M.M.S.; supervision, M.M.S. and M.V.; project administration, K.M.G.M. and M.M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and its later amendments, and approved by the Institutional Review Board (or Ethics Committee) of the University of Pretoria (691/2019)

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We would like to acknowledge the support staff including but not limited to the Nuclear Medicine Technologists, Administration staff and the staff in the departments of radiation oncology and pathology.

**Conflicts of Interest:** The authors declare no conflict of interest.
