Position emission tomography (PET) is an imaging modality that uses a tomographic technique to measure the 3-dimensional distribution of positron-emitting labeled radiotracers. 18Flourine-labeled glucose (18-fluoro-2-deoxyglucose; 18F-FDG) is the most commonly used tracer for PET studies. Most cancers characteristically have increased utilization of glucose, upregulation of glucose transporters, and consequently an increased uptake of glucose analogue FDG on PET studies. This allows for the noninvasive quantification of tumor metabolic activity, measured by the standardized uptake value (SUV) of FDG by the tumor compared with the FDG uptake value of background tissues. The addition of computed tomography (CT) allows the morphological evaluation of the tumor to improve the accuracy of the size and localization, combined with the metabolic measurement on the PET images (1).
One advantage of 18F-FDG PET/CT is its ability to detect metabolic changes before anatomical changes occur, making it more sensitive to the nodal disease status of patients (2, 3). In the developed world, several studies have assessed the role of 18F-FDG PET/CT in the staging and management of patients with locally advanced cervical cancer (LACC), and results from 18F-FDG PET/CT scans can provide important prognostic information for these patients with cervical cancer (4). The staging of cervical cancer is based on the FIGO (International Federation of Gynecology and Obstetrics) classification, which is a clinical staging system involving a thorough clinical examination and anatomical imaging. In a resource-constrained setting, a chest x-ray, abdominal ultrasonography, and cystoscopy are routinely performed on patients with cervical cancer with further investigations considered on a case-by-case basis, and depending on resources and availability (5).
Morkel et al. reported that the addition of 18F-FDG PET/CT to the initial screening protocols for patients with FIGO stage IIIB cervical cancer revealed relevant disease information that was missed by clinical staging and other imaging techniques. The information gathered from the 18F-FDG PET/CT resulted in a change in treatment protocols in a significant number of patients with cervical cancer (5). In a study by Kidd et al., 53% of the 234 patients with cervical cancer (FIGO stage IB1–IVA), had FDG-avid lymph nodes on 18F-FDG PET/CT scans, with nodes visualized as high as the supraclavicular region. Lymph node status as visualized on the 18F-FDG PET/CT scans had a significant effect on patient outcomes, and the presence and number of lymph node metastases are known prognostic factors (6, 7). The sensitivity of the PET/CT studies depends on the stage of disease, with limited benefit of the use of 18F-FDG PET/CT scans seen in early stages of cervical cancer (8-10). The typical metastatic pattern for cervical cancer is from the nodes directly draining the cervix (parametrium and paracervix), to the external iliac, internal iliac, obturator, and mesorectum groups (sometimes referred to as level one), to the common iliac and presacral groups (or level two groups), to para-aortic nodes (level three) (11-13).
A well-known challenge regarding the interpretation of 18F-FDG PET/CT scans is the risk of false-positive results in patients with acute infections or inflammatory conditions (14). Activated inflammatory cells, such as macrophages, lymphocytes, neutrophil granulocytes, and fibroblasts, have been shown to avidly take up FDG, resulting in an FDG uptake that is higher than the background uptake (15). The gradual decrease in CD4 cells of human immunodeficiency virus (HIV)-1-infected individuals is associated with a decline in the function of patient’s immune system. The homing theory suggests that the homing of HIV-infected CD4 cells from the blood to the lymph nodes where they undergo apoptosis is a possible cause for the decrease in CD4 counts (16, 17). This correlates with generalized lymphadenopathy typically seen in HIV-infected patients.
Studies have shown the importance of taking the HIV status of the patient into account when interpreting the results of 18F-FDG PET/CT images of patients with cancer. Morkel et al. found a significant inverse relationship between the CD4 count of HIV-positive patients and the mean SUV of their lymph nodes (5), and another study found that HIV-positive patients with high viral load levels may have benign hypermetabolic foci visualized on 18F-FDG PET/CT images (18). This may result in false-positive interpretations of malignancy. Increased FDG uptake of the lymphoid tissue is seen in the head and neck region during acute HIV infection, in the peripheral nodes (including the axillary and inguinal nodes) during chronic infection, and in the abdominal lymph nodes during late stages of HIV infection (18-20). A detailed study by Brust et al. showed that healthy HIV-infected individuals with suppressed viral loads, and their HIV-negative counterparts, had little to no FDG uptake in lymph nodes; however, HIV-positive patients with high viral loads and advanced HIV infection had increased FDG uptake in the peripheral lymph nodes (21).
To the best of our knowledge, there are no larger studies investigating the nodal characteristics and distribution patterns of HIV-positive, compared to HIV-negative, patients with LACC in a resource-constrained setting. In this paper, we report on the results of 18F-FDG PET/CT images from participants enrolled in an ongoing phase III randomized controlled trial on the effects of the addition of modulated electro-hyperthermia to chemoradiation protocols in South Africa. We aim to identify differences in the characteristics of positive lymph nodes between HIV-positive and HIV-negative LACC participants. This may assist in more accurate reporting and staging of cervical cancer on 18F-FDG PET/CT studies in geographic areas with a high incidence of HIV infection.
Materials and Methods
The data for this analysis are part of a phase III randomized controlled trial investigating the addition of modulated electrohyperthermia to chemoradiotherapy for the treatment of patients with LACC. The trial is being conducted in our local academic hospital by the Department of Radiation Sciences of the University of the Witwatersrand, Johannesburg, South Africa. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional/national research committee (Human Research Ethics Committee approval: M120477) and the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The trial was registered on the South African National Clinical Trials Register before recruitment started (ID: 3012) and at ClincialTrials.gov ID: NCT03332069. Informed consent was obtained from all individual participants included in the study.
As part of the screening evaluation for the trial protocol, potential participants had additional hematological investigations including HIV tests and CD4 count, and they underwent 18F-FDG PET/CT imaging studies before randomization, but only after signing the informed consent.
Patients were considered eligible for screening for the trial and were therefore found eligible for this analysis if they were>18 years old, with histologically confirmed squamous cell carcinoma of the cervix, FIGO stage IIB (with distal involvement of the parametrium) to IIIB (staged clinically), creatinine clearance of >60 mL/min, reported knowing their HIV-status, did not have any AIDS-defining illnesses except for cervical cancer, were deemed fit for treatment with radical intent, and had a life expectancy of >12 months.
Once the screening process was complete, exclusion criteria for randomization for the trial based on the screening results included a CD4 count of <200 cells/µL if HIV-positive, visceral metastatic disease, bilateral hydronephrosis, or fistulas visualized on 18F-FDG PET/CT scans.
In total, 270 participants were screened, 250 of whom underwent pretreatment 18F-FDG PET/CT imaging studies, and 210 were deemed eligible for trial and were randomized for treatment. For the purpose of this report, we evaluated all 250 of the 18F-FDG PET/CT imaging studies conducted as part of the screening process.
Patient Preparation and Imaging Protocol
Participants were prepared for the 18F-FDG PET/CT scans according to department standard protocols. Plasma glucose was checked before the scan, and hydration was administered to participants whose plasma glucose levels were ≥11 mmol/L or more to lower glucose levels and minimize the interference by high glucose (1). The PET/CT study was rescheduled if the plasma glucose could not be adequately lowered. Our protocol aimed for an uptake period of ±1 h between injection and scanning (1) depending on the logistics. Participants were injected with a mean of 378.51 MBq (minimum, 135.79 MBq; maximum, 462.5 MBq) of FDG. The participants’ mean plasma glucose level was 5.7 mmol/L (minimum, 3.7 mmol/L; maximum, 12.2 mmol/L), and there was a mean interval of 66.5 min (minimum, 50; maximum, 125) between injection time and acquisition of body image.
The scanning protocol begins with a scout image, followed by a low-dose CT image (with intravenous contrast when not contraindicated) of the whole body. This was followed by “whole-body” PET imaging in 7 bed positions, 3 min per bed, starting from mid-thigh to the base of the skull. A diagnostic CT of the chest was then performed. The 18F-FDG PET/CT scans were obtained using a Siemens Biograph™ 40 PET/CT (Siemens Healthcare GmbH; Erlangen, Germany).
The nodal characteristics assessed included number of nodes and their largest diameter on CT scans, uptake assessed by measured maximum SUV (SUVmax), symmetry, and relationship between nodal size and SUVmax uptake. SUVmax ≥ 2.5 was used as a cutoff for active nodes. Patterns were compared between different anatomical regions that included the pelvis, abdomen, thorax, and head and neck. First, the characteristics of all of the nodes were analyzed depending on whether the node appeared in the HIV-negative or -positive participants (nodal analysis). Then the participants in each group (HIV-positive and HIV-negative) were assessed, and the characteristics of their nodes were compared (participant analysis). A subgroup analysis of specific node groups responsible for the drainage of the cervix and nodes known to be positive in HIV infection was conducted.
Frequencies and percentages were calculated, and we used chi-square frequency tables, unpaired t test, linear regression and logistic regression models to measure associations, differences, and odds ratios. STATA 13.0 Statistics software program (Stata Corporation, College Station, TX) was used to analyze the data.
This study was funded by the National Research Foundation of South Africa (TP12082710852). 18F-FDG was supplied at a discounted rate by Radioisotopes SOC Ltd. NTP Radioisotopes SOC Ltd is a state owned company and a subsidiary of the South African Nuclear Energy Corporation. The company manufactures and supplies radioisotopes. Funding sources had no role in study design, implementation, data collection, and data analysis or development/review of the manuscript.
Participant characteristics are listed in Table 1. Majority of the participants were African (92%); 128 (51%) were HIV-positive, 119 (82%) of whom were on antiretroviral therapy (ART) at the time of the scan. The median CD4 count was 506.5 cells/µL (range, 95–1524 cells/µL), and 10 participants had a CD4 count <200 cells/µL.
i] Abbreviations: BMI: Body Mass Index; HIV: Human Immunodeficiency Virus; ECOG: Eastern Cooperative Oncology Group.
Overall Nodal Analysis
In total, 2299 nodes were visualized, 1314 of which had SUVmax ≥ 2.5 and were included in the analysis (HIV-positive, 713 [54%]; HIV-negative, 601[46%]). A summary of the number of nodes per region, and of the HIV status of the participant in which the node was visualized, can be seen in Table 2.
|Head and Neck||225||139||[62%]||86||[38%]|
|Total number of nodes||1314||713||[54%]||601||[46%]|
SUVmax Analysis by Region.
Overall, the mean SUVmax was significantly higher in the pelvis and the abdomen compared with the head and neck region (P < .001) and thorax (P < .001). However the difference in SUVmax between the abdominal and pelvic nodes was not significant (P = .397). The mean SUVmax was the highest in the pelvis, and this was significantly higher than that in any other region (P < .001).
SUV and HIV Status.
When considering all nodes, in a multivariate logistic regression analysis, a higher SUVmax was a predictor of the HIV status of the participants in which the node was visualized (OR, 0.97; P = .022; 95% CI, 0.945–0.996). However when considering the nodes by region, the abdomen was the only region in which the SUVmax of the nodes was predictive of the HIV status of the participants (OR, 0.92; P = .006; 95% CI, 0.86–0.98). When considering the SUVmax of all of the visualized lymph nodes, grouped into bins of 10 starting from 3, HIV-negative participants were more likely to have SUVmax of between 3 and 23 and SUVmax > 23 is associated with an HIV-positive status (chi-square: P = .028). The association was however not significant in a chi-square analysis of each region. In HIV-positive participants, the relationship between SUVmax and CD4 count was not significant (P = .131; 95% CI, −0.003 to 0.0003).
Size Analysis by Region.
The largest nodes were seen in the pelvis, and the difference in sizes between the pelvic and abdominal nodes was significant (P < .0001). The lymph nodes in the abdomen were significantly larger than the nodes of the thorax (P = .0047) and those in the head and neck (P < .0001).
Size and HIV Status.
When considering the size of the nodes (in centimeters), grouped into 0.5-cm bins, the only significant finding was that the nodes from the abdomen of HIV-positive participants are more likely to be <1.49 cm (106 out of 146 vs 80 out of 142), and the nodes from HIV-negative participants are more likely to be 1.49 cm–2.49 cm (51 out of 145 vs 28 out of 142; chi-square: P = .027). In a logistic regression analysis, lymph node size was not predictive of the HIV status; however, the increased size was significantly associated with reduced CD4 counts in HIV-positive participants (coefficient, –21.34; P = .033; 95% CI, 40.94 to 1.75): Supplemental Image 3 is of the participant (42) who was HIV-positive and had the highest SUV measured in a lymph node (54.2).
Relationship Between SUV and Size.
The mean size of the visualized nodes was 1.55 cm (standard deviation [SD], 0.93; minimum, 0.5 cm; maximum, 14.2 cm chain of nodes) and the mean SUVmax was 6.43 (SD, 4.87; minimum, 2.5; maximum, 54.21). There is a moderately strong (r = 0.5) and significant (P < .001) positive relationship between the overall SUVmax and node diameter. This relationship was significant for each region, regardless of the HIV status, when analyzed independent of the participants and when analyzed per participant (Figure 1).
Analysis of Lymph Nodes Within Participants (Participant Analysis)
In paired t-test analyses, the maximum size of the nodes, the maximum SUVmax of the nodes, and the number of nodes overall and by region were assessed in each participant. There were no significant differences in the numbers of nodes in HIV-positive participants compared with HIV-negative participants (P > .05). These results are summarized in Table 3 and represented graphically in online supplemental Images 1, 2, and 4. To determine if there is any association between HIV status and a specific size or SUVmax, nodes were organized into bins and frequency tables were generated. No significant association was seen when looking at participants’ maximum or mean size or SUVmax. The only significant association between HIV status and number of lymph nodes in participant was seen in the head and neck region, where >4 nodes are more strongly associated with HIV infection (chi-square: P = .004).
A subgroup analysis of specific lymph node groups known to be positive on 18F-FDG PET/CT in HIV-positive patients to determine whether, in our sample, there is a risk of understaging of the cervical cancer by attributing positive nodes to HIV-infection rather than malignancy was conducted.
Head and Neck Nodes (Known to Be Positive in Early HIV Infection).
Of specific interest in the head and neck region was the significantly higher number of supraclavicular nodes visualized in HIV-positive participants (mean, 0.15; SD, 0.454; 95% CI, 0.069–0.228) compared with the number of those in HIV-negative participants (mean, 0.05; SD, 0.217; 95% CI, 0.01–0.09) (P = .0296). Posterior cervical nodes were also visualized significantly more (P = .0037) in HIV-positive participants (mean, 0.21; SD, 0.053; 95% CI, 0.107–0.315) than in HIV-negative participants (mean, 0.04; SD, 0.022; 95% CI, 0.002–0.084). Online supplemental Images 5a and 5b are of the HIV-positive (5a) and HIV-negative (5b) participants with the highest number of lymph nodes visualized in the head and neck region.
Axillary Nodes (Known to Be Positive in Chronic HIV Infection).
Seven out of 122 (6%) HIV-negative and 29 out of 128 (23%) HIV-positive participants presented with axillary nodes visualized on 18F-FDG PET/CT scans. In a logistic regression analysis, each increase in the number of axillary nodes is associated with a 2.7-fold increase in the odds of the participant being HIV-positive (P = .001), with HIV-positive participants having a maximum of 3 axillary nodes and a mean of 0.37 nodes, and HIV-negative participants having a maximum of 2 and a mean of 0.08 nodes (P = .001). Symmetry was not associated with HIV status (symmetry of axillary nodes in HIV-positive participants, 17 out of 29, 59%; symmetry of axillary nodes in HIV-negative participants, 3 out of 7, 43%; P = .451).
Inguinal Nodes (Known to Be Positive in Chronic HIV Infection).
The SUVmax and size were not predictors of HIV status in the inguinal lymph nodes. However the number of inguinal lymph nodes and the symmetry was significantly associated with HIV infection (symmetry of inguinal nodes seen in HIV-positive participants, 11 out of 21, 52%; symmetry of inguinal nodes seen in HIV-negative participants, 2 out of 14, 14%; P = .022). The number of nodes was a significant predictor of HIV status (P = .047; 95%CI, 1.01–2.83; OR, 1.69) in a logistic regression and there was a significant difference in the mean number of nodes between the 2 groups with a mean of 0.6 nodes in the HIV-positive participants and a mean of 0.3 nodes in the HIV-negative participants (t-test: P = .0417).
Colic or Mesenteric Nodes (Known to Be Positive in Late-Stage HIV Infection).
There were no significant associations between HIV status and the number, symmetry, SUVmax, or size of the colic, paracolic, or mesenteric lymph nodes.
Nodes Involved in Draining the Cervix.
Nodes included in this analysis are paracervical, parametrial, presacral, sacral, external and internal iliac, common iliac, hypogastric, and para-aortic nodes. There were no significant relationships between HIV status and SUVmax, size, number, or symmetry in these nodes. Online supplemental Images 6a and 6b are of the HIV-positive (6a) and HIV-negative (6b) participants with the highest number of lymph nodes visualized in the pelvic region.
In our sample, aortopulmonary nodes were significantly (P = .0096) more frequent in HIV-negative participants (mean, 0.15; SD, 0.454; 95%CI, 0.01–0.08) than in HIV-positive participants (mean, 0.05; SD, 0.217; 95%CI, 0.08–0.22), and abdominal caval nodes had higher SUVmax in HIV-negative participants (mean, 7.941) than in HIV-positive participants (mean, 5.685; P = .0430). Online supplemental Image 7 is of the HIV-positive participant with the largest node (3.1 cm) and the node with the highest SUV (24.38) in the head and neck region and in the thorax (SUV: 21.77; size: chain of >14 cm).
The uniformity in presentation of nodes draining the cervix in HIV-positive and HIV-negative LACC participants confirms that visualization of these nodes on 18F-FDG PET/CT scans may be attributed to malignancy rather than HIV status of the participants in our sample. Overall there are some novel differences between the visualization of lymph nodes in HIV-positive and HIV-negative participants in our study, which may be of use in deciding the likely cause of the increased FDG uptake in some patients; however, the overall pattern of presentation and the nodal characteristics are similar between the two groups.
There is a significant association between higher SUVmax and larger lymph node diameter visualized in 18F-FDG PET/CT. SUVmax and node size were consistently higher in the pelvis and abdomen than in the thorax and head and neck regions regardless of the HIV status. The number of nodes visualized in the pelvis was higher that than in any other region. These findings are all in line with the literature and the expected route of metastatic spread of cervical cancer (11-13). In a report by Iyengar et al. (22) on the FDG uptake in 12 patients with acute HIV infection and 11 patients with mid/chronic HIV infection, all patients presented with an FDG uptake that was greater in the cervical and axillary lymph nodes than in the inguinal and iliac chains. The patients in their study were however not all on ART, and there was a significant correlation between FDG PET signal and viral load (22). In our sample of HIV-positive cervical cancer participants, most of whom were on ART, the highest SUVmax was still seen in the pelvis and abdomen.
There is a risk of underestimating the extent of lymph node involvement in HIV-positive participants if the increased FDG uptake of a node is incorrectly attributed to HIV infection. We were therefore specifically interested in the differences in lymph node presentation between HIV-positive and HIV-negative participants in our sample.
The visualization of head and neck nodes is frequently cited in the literature as being associated with acute HIV infection (18-20, 22). The majority of participants in our study (82%) were on ART for >6 months, and the majority of these not on ART were aware of their HIV-status and were not likely to be in acute stages of HIV infection. The majority of our participants also had a CD4 count >200 cells/µL. Although more head and neck lymph nodes were visualized in the HIV-positive participants in our sample, differences in the number of head and neck nodes visualized between HIV-positive and HIV-negative participants were not significant. However, >4 nodes in the head and neck region were more likely to be seen in HIV-positive patients. Further histological evaluation of nodes in this region for malignancy in HIV-positive patients may therefore still be warranted.
Kidd et al. (6), reported that 53% of 234 patients with LACC presented with FDG-avid lymph nodes on 18F-FDG PET/CT scans, and in that sample, the nodes were visualized as high as the supraclavicular region. Our sample only included participants in FIGO stages IIB to IIIB (staged clinically), and the number of participants with FDG-avid nodes in our study was considerably higher (90%), likely because of the higher proportion of our participants with more advanced stages of disease. We also saw increased FDG uptake in the supraclavicular nodes that are known to be potential sites of metastatic spread in patients with cervical cancer. In our sample however, the number of supraclavicular nodes visualized was significantly higher in the HIV-positive participants despite the management of the HIV infection with ART. Based on these findings, further assessment of supraclavicular nodes visualized in HIV-positive cervical cancer patients to rule out HIV infection as a cause of the increased FDG-uptake may be warranted.
Although axillary nodes were significantly associated with HIV infection in our sample, the relationship was not exclusive, with 6% of HIV-negative participants presenting with FDG-avid axillary nodes. The spread to axillary nodes can therefore not be conclusively excluded in HIV-positive immunocompetent patients with chronic HIV-infection who are on ART and have a CD4 count>200 cells/µL.
Symmetrical inguinal lymph nodes are more likely to be visualized in HIV-positive participants, even if their HIV infection is managed. This is in line with the literature that has found inguinal nodes to be FDG-avid during chronic HIV infection.
In HIV-positive participants, larger nodes were associated with lower CD4 counts and the only difference in size between HIV-positive and -negative participants was seen in the abdomen, where larger nodes were more common in HIV-positive participants. The literature shows a significant association between CD4 count and SUVmax (16, 17); however, we did not see the same significant association. The majority of our participants had a CD4 count >200 cells/µL and the association may therefore be more relevant to patients with a CD4 count<200 cells/µL. Based on our results it appears that the SUVmax in immunocompetent patients is a reliable measurement of metabolic activity in potentially metastatic nodes.
Fewer participants in this study were in the late stage of HIV infection, with only 7.8% of HIV-positive participants presenting with a CD4 count <200 cells/µL. The lack of colic or mesenteric nodes visualized in our sample is therefore in line with the literature (19).
Our results indicated that 18F-FDG PET/CT is a reliable staging method for LACC regardless of the HIV status, provided the patients are not in acute stages of HIV infection, have a CD4 count >200 cells/µL, and/or are on ART. The results of this study are clinically relevant, as this may be of assistance in determining which lymph nodes visualized on 18F-FDG PET/CT are likely to be due to metastatic spread in HIV-positive patients with LACC, and in preventing understaging of HIV-positive LACC patients.