False negative rate at 18F-FDG PET/CT in para-aortic lymphnode involvement in patients with locally advanced cervical cancer: impact of PET technology

Background The identification of factors responsible for false negative (FN) rate at 18F- Fluorodeoxyglucose (FDG) Positron Emission Tomography /Computed Tomography (PET/CT) in para-aortic (PA) lymph nodes in the presurgical staging of patients with locally advanced cervical cancer (LACC) is challenging. The aim of this study was to evaluate the impact of PET/CT technology. Methods A total of 240 consecutive patients with LACC (International Federation of Gynecology and Obstetrics, FIGO, stage IB2-IVA) and negative Magnetic Resonance Imaging (MRI) and/or Computed Tomography (CT) and negative 18F-FDG PET/CT in the PA region, undergoing laparoscopic PA lymphadenectomy before chemoradiotherapy were included. The FN rate in patients studied with Time of flight (TOF) PET/CT (TOF PET) or non-Time of flight PET/CT (no-TOF PET) technology was retrospectively compared. Results Patients presented with FIGO stage IB (n = 78), stage IIA-B (n = 134), stage III (n = 18) and stage IVa (n = 10), squamous cell carcinoma (n = 191) and adenocarcinoma (n = 49). 141/240 patients were evaluated with no-TOF PET/CT and 99/240 with TOF PET/CT. Twenty-two patients (9%) had PA nodal involvement at histological analysis and considered PET/CT FN findings. The FN rate was 8.5% for no-TOF PET and 10% for TOF PET subgroup respectively (p = 0.98). Ninety patients (38%) presented with pelvic node uptakes at PET/CT. The FN rate in the PA region was 18% (16/90) and 4% (6/150) in patients with and without pelvic node involvement at PET/CT respectively (19 vs 3% for no-TOF PET and 17 vs 5% for TOF PET subgroup). Conclusions In LACC, FN rate in PA lymph nodes detection is a clinical issue even for modern PET/CT, especially in patients with pelvic uptake. Surgical lymphadenectomy should be performed in case of negative PET/CT at PA level in these patients, while it could be discussed in the absence of pelvic uptake.


Background
The therapeutic approach in locally advanced cervical cancer (LACC) is a challenge and based on tumor volume, nodal metastasis, and clinical stage [1]. In LACC (i.e. FIGO stage IB2 to IVA) chemoradiation in association with pelvic External Beam Radiation Therapy (EBRT) is recommended in case of metastatic pelvic but negative Para-Aortic (PA) lymph nodes; in case of metastatic PA lymph nodes, an extended radiation field to PA region is applied [1,2].
Therefore, imaging techniques with high diagnostic accuracy are fundamental for disease staging. Diagnostic imaging in cervical cancer includes abdomino-pelvic Magnetic Resonance Imaging (MRI), Computed Tomography (CT) scan and 18 F-Fluorodeoxyglucose (FDG) Positron Emission Tomography /Computed Tomography (PET/CT) 18 F-FDG PET/CT [3][4][5][6]. 18 F-FDG PET/CT may help to rule out loco-regional lymph nodes and distant metastasis through a wholebody examination. Indeed, recent guidelines recommend PET/CT over CT in initial staging of LACC to detect extra pelvic disease and for EBRT plan delineation [1]. However, in patients with LACC, high false negative rate for PET/CT in detecting metastatic PA lymph nodes has been reported in several published studies [7][8][9][10][11]. For this reason, surgical staging is still considered in patients with negative presurgical 18 F-FDG PET/CT at PA level [12]. During the last years, we assisted to a rapid development of PET technology. In particular, the introduction of Time of Flight (TOF) technique using crystal materials with relatively high time resolution should theoretically increase the sensitivity and consequently lesion detectability [13][14][15].
The aim of this study is to determine the impact of TOF PET/CT vs no-TOF PET/CT technology on false negative rate in PA lymph node detection in the presurgical staging of patients with LACC.

Study design and patient enrolment
All consecutive patients with locally advanced cervical cancer (LACC) according to FIGO classification (stage IB2-IVA) treated between 2007 and 2015 at Gustave Roussy (GR), with negative morphological imaging (abdomino-pelvic MRI in most cases or CT scan and pelvic MRI) and negative 18 F-FDG PET/CT in the paraaortic area (PA), undergoing laparoscopic PA lymphadenectomy before chemoradiotherapy were included. Patients with a poor prognosis according to histological subtype or peritoneal carcinomatosis were excluded. Institutional review board of Gustave Roussy approved the study and waived the need to obtain informed consent.
The surgery included the removal of PA nodes from the aortic bifurcation to the left renal vein. PA above and below the inferior mesenteric artery, preaortic, superficial intercavoaortic and precaval groups were removed. Pelvic nodes were not resected because included in the radiotherapy field.
All patients were treated with pelvic external beam radiation therapy (45 to 50 Gy) associated to a concomitant cisplatin-based chemotherapy protocol after surgery. Patients with histologically proven PA node metastasis after staging laparoscopic surgery were treated with pelvic and extended to PA region EBRT (45 to 60 Gy) with concomitant cisplatin chemotherapy.
A total of 258 patients were initially evaluated. Eighteen patients were excluded from analysis because of a final IB1 staging with involved pelvic nodes (n = 9), small cell carcinoma histology (n = 1), discovery of peritoneal carcinomatosis during laparoscopy (n = 5), ovarian metastasis (n = 1) and dubious findings for lymph node metastases in the para-aortic region at PET/CT (n = 2) (Fig. 1). In the end, a total of 240 patients were considered for the analysis and two subgroups of patients were retrospectively identified according to the PET/CT technology used: patients studied with TOF PET/CT and patients studied with no-TOF PET/CT. 18

F-FDG pet/CT
All the patients underwent a 18 F-FDG PET/CT before surgery. One hundred and eighteen patients underwent 18 F-FDG PET/CT at Gustave Roussy and the remaining 122 in others different Nuclear Medicine Services in France according to guidelines [16]. After 6 h of fasting, patients performed images acquisition 60 min after the administration of a median activity of 254 MBq of 18 F-FDG (range: 137-591 MBq). Median glycaemia before PET/CT was 4.6 mmol/L (range:0.74-10.4).
On the basis of constructor characteristics and reconstruction algorithms, PET/CT scanners were divided into two categories: TOF PET/CT (TOF PET) using 3D coincidence algorithms for image reconstruction and including TOF system (PET/CT Gemini-TF Philips, PET/ CT Biograph mCT20-mCT40 Siemens, PET/CT Discovery 690 General Electric Healthcare, PET/CT Discovery 710 General Electric Healthcare) and no-TOF PET/CT (no-TOF PET) with 2D-3D coincidence algorithms for image reconstruction and not provided with TOF system (PET/CT Biograph 6-Siemens, PET/CT Gemini GXL).

Statistical analyses
Patient characteristics were reported as relative/absolute frequencies for categorical covariates and as median (range) for continuous ones. Their potential differences, when stratified as TOF-PET vs. no-TOF PET, were inferred by the Fisher's exact test for categorical variables and the Mann-Whitney test for continuous ones. All reported p-values were obtained by the two-sided exact method, at the conventional 5% significance level. Data were analysed by R 3.6.3 (R Foundation for Statistical Computing, Vienna-A, http://www.R-project.org).

Patient characteristics
The median age was 45 years  Before surgery, 18 F-FDG PET showed uptake in pelvic lymph nodes in 90/240 (38%) patients. The uptake was unilateral in 57/90 (63%) patients, located in the right (45%) and in the left region (51%) respectively (3.5%   Unilateral pelvic uptake 57 63 Bilateral pelvic uptake 33 37 Delay in days between PET-surgery, median (range) 15.5  No of para-aortic nodes removed, median (range) 15  No of patients with positive para-aortic nodes 22 9 Size of the biggest para-aortic nodes involved, mm < 5 9 4 1

> 5 13 59
Abbreviations: FIGO International Federation of Gynecology and Obstetrics, BMI Body Mass Index, PET Positron Emission Tomography missing data). All patients with unilateral uptake presented with ≤2 positive lymph nodes at 18 FDG PET/CT. Pelvic uptake was bilateral in 33/90 (37%) of patients (Fig. 2). Thirty patients presented with ≤2 positive lymph nodes at 18 FDG PET/CT by side, while the remaining 3 three patients presented with > 2 positive lymph nodes (5 + 2 in 1 case, 2 + 3 in 1 case and 1 + 3 in 1 case in the left and right region respectively). In the remaining 150/ 240 (62%) there was no uptake in the pelvic region. In regard to surgical laparoscopy techniques, 235 patients underwent a retroperitoneal approach and five a transperitoneal approach. One patient had conversion to laparotomy. The mean delay between 18 (Table 3).

Discussion
PA lymph node metastases occur in 10-25% of patients with LACC and are correlated with pelvic, common iliac nodes involvement and larger primary tumor size [17,18]. The detection of metastatic PA lymph nodes is a fundamental step to define the correct therapeutic approach in LACC and to improve patient outcome [19][20][21][22]. False negative rates for 18 F-FDG PET and PET/CT in PA areas have been reported in several studies ranging from 5 to 17% [7,11]. The identification of factors responsible for false negative PET/CT findings in PA region in LACC is challenging. In the first studies on this topic, false negative rate was mainly related to lymph node size < 5 mm considered beyond the machine resolution. In the study by Roh et al. including patients with stages IA to IVA PET sensitivity was 38% increasing to 52% in case of lymph nodes > 5 mm and 65% in the case of lymph nodes > 10 mm [7].
The use of hybrid machines such as PET/CT has greatly improved the accuracy of the technique allowing a better localisation and characterisation of abdominal FDG uptakes [23]. However, results among the studies are variable and even in more recent studies false negative rate of 18 F-FDG PET/CT remains high up to 22%, in particular in early-stage disease [24][25][26][27]. The retrospective study by Leblanc et al. including 125 patients with cervical cancer in stage IB2 -IIA with negative preoperative CT scans or MRI, FDG PET or PET/CT showed a sensitivity of 33% in PA lymph nodes detection with false negative rate of 67% [10]. A recent metanalysis including 29 studies focused on 18 F-FDG PET alone and PET/CT reported a pooled sensitivity of 0.72 in detecting both pelvic and PA lymph nodes, 0.83 in case of locally advanced disease and 0.41 in early-stage disease [6]. Another recent recent metanalysis from Yu et al. including 14 studies evaluated specifically the diagnostic performance of PET/CT for the detection of PA lymph nodes [28]. The authors reported a pooled sensitivity and a pooled specificity of 0.71 (95% CI: 0.54-0.83) and 0.97 (95% CI: 0.93-0.98) respectively and a pooled positive likelihood ratio (PLR) and negative likelihood ratio NLR (NLR) of 21.53 and 0.30, respectively. The authors concluded that 18 F-FDG PET is an important imaging method for the detection of PA lymph nodes but underline the limitations of the included studies, mostly retrospective and with low sample size. The ACRIN6671 trial reported a sensitivity for 18 F-FDG PET with contrast enhanced CT of 0.83 in the detection of pelvic lymph nodes decreasing to 0.50 for PA lymph nodes [29].
In a previous multicentric prospective study evaluating 237 patients in the same setting, our group demonstrated 18 F-FDG PET/CT false negative rate of 12% and the results supported the conclusion that PA lymph node dissection should be performed in LACC in case of negative 18 F-FDG PET/CT in PA region for a proper patient staging [11].
The impact of PET device has never been taken into account in order to evaluate and to compare accuracy in this scenario. The aim of this study was to evaluate the impact of PET/CT technology on false negative rate in PA lymph node detection in LACC staging. In the last decades there have been multiple advances in PET technology with improvement in cancer imaging [30]. The Table 3 False negative rate of 18 F-FDG PET/CT in para-aortic region according to pelvic uptake  [31,32]. This technology is useful in clinical practice, allowing shorter examinations, reduction of the radiation dose to the patient, successful scanning of overweight patients, clearer characterisation of low uptake areas and visualization of smaller lesions [31,32]. Our data show that the false negative rate in PA lymph node detection in patients evaluated by PET/CT scan without TOF system integration compared to PET/CT with TOF integration is comparable (8.5 vs 10%; p = 0.98). These findings suggest that even if the sensitivity or lesion detectability is supposed to be higher in modern PET/CT scanner the limitation of the technique to assess PA lymph node is still a clinical issue in this specific clinical setting. The resolution of the PET/CT technology with TOF is considered between 4 and 6 mm [13,31,32]. Considering lesion size, in our study 59% of the metastatic lymph nodes detected at histology presented with a diameter > 5 mm with a similar distribution in the two groups. In our opinion, the size cannot be considered the only cause of the false negative rates of 18 F-FDG PET/CT. Some studies suggest that 18 F-FDG uptake can be influenced by tumoral histotype in primary and metastatic lesions, with higher 18 F-FDG uptake in squamous cell compared to non-squamous cell tumors. In the study of Lin et al. nodal metastasis was less likely to be detected by 18 F-FDG-PET in patients with earlystage adenocarcinoma compared to squamous carcinoma [33]. Nevertheless, in our study almost 80% of the patients and 21/22 patients with false negative 18 F FDG PET/CT presented with squamous cell carcinoma. The para-aortic lymph nodes are close to the urinary tract and 18 F-FDG physiological urinary elimination can hide small 18 F-FDG avid lesions. In this case, the use of hybrid imaging such as PET/MRI and contrast enhanced diagnostic CT coupled with PET might better differentiate between pathological and physiological uptake, but well-designed trials are warranted to confirm this assumption [34]. Patient BMI should be also taken into account because in overweight patients the quality of images can be affected and the detection rate reduced, but the BMI in the two groups of our study was similar.
Finally, patients with pelvic pathological nodes metastases, compared to patients without pathological pelvic nodes, have a higher risk to develop PA metastases [17,18]. In our study the percentage of pelvic positive lymph nodes in TOF PET compared to no-TOF PET group (42% vs 34%) was not statistically different (p = 0.5). FN rate in PA region was higher in case of pelvic uptake at 18 F-FDG PET/CT, especially when bilateral, compared to negative pelvic uptake cases. These data are confirmed in both no-TOF and TOF PET groups with FN rate of 19% vs 17 and 3% vs 5% respectively according to pelvic vs no pelvic uptake status.
A very recent retrospective study (ONCO-GF) reported a PET/CT sensitivity in identifying PA lymph nodes metastases of 23.5%. Among 151 patients with negative PA lymph node at PET/CT, 26 (17,2%) had histological proven metastases including 21 with macroscopic lesions and the percentage increased considering patients with positive pelvic uptake (18,1%) confirming our observation. The authors conclude that para-aortic surgical staging contributes significantly to individualise the radiation treatment plan especially in patients with positive pelvic uptake [35]. FRANCOGYN study including 647 patients with LACC treated by concurrent chemoradiation therapy (CRT) and no evidence of paraaortic metastasis on pre-operative imaging work-up, confirmed in multivariate model analysis surgical staging as an independent prognostic factor for Disease Free Survival [36]. On the other hand, a retrospective study from Mayo Clinic registry including 148 patients did not find differences comparing surgical versus radiological staging (mainly based on CT, MRI and PET combination) in 5-year PFS (HR 1.11, 95% CI 0.54-2.30, p = 0.77) and 5-year OS (HR 1.02, 95% CI 0.46-2.29, p = 0.96). The presence of PA lymph nodes metastasis significantly predicted unfavourable PFS (HR 2.76, 95% CI 1.23-6.18, p = 0.01) and OS (HR 3.46, 95% CI 1.40-8.55 [37]. More consistent data will be forthcoming in large trial of surgical vs radiological staging. A prospective study, including 600 patients with stage IB2-IVA cervical cancer, is ongoing in order to evaluate the impact on survival of standard chemoradiotherapy treatment based only on 18 F-FDG PET/CT staging compared to an adapted treatment based on PA lymph node dissection staging [38]. However, considering our results and previous published data, surgical lymph node dissection could be discussed in patients without involved pelvic lymph nodes on 18 F-FDG PET/CT, considering the very low false negative rate in this case. On the other hand, in case of 18 F-FDG PET/CT uptake in the pelvic region, in particular if bilateral, the risk of FN findings in PA lymph node is higher and surgical dissection could be preferred.
Our study presents several limitations, in particular the retrospective nature, the heterogeneity of the PET/ CT scanners employed and the lack of a standardized protocol for images acquisition. Furthermore, for strength conclusions on the impact of TOF technique a direct comparison of images derived from the same patient with and without TOF reconstruction method should be performed through a proper trial. On the other hand, the impact of PET/CT device has been tested in a real clinical setting, in a wide sample size population with histological confirmation, and in comparable patient populations. At this moment the increasing availability of Digital PET scanner will probably change the scenario and a direct comparison among different reconstruction methods will be the next step to evaluate the impact of technological development in this clinical setting. Finally, patients with positive PA at 18 F-FDG PET/CT have not been included in the study and the analysis of true or false positive findings in the PA region has not been performed because a reliable confirmation by histology would not have been available being the patient directly sent to extended EBRT protocol.

Conclusions
The FN rate in PA lymph nodes detection in LACC is high, independently from PET/CT technology. Surgical staging should be preferred in case of negative presurgical imaging in the PA region, in particular in case of 18 F-FDG PET/CT abnormal uptakes in pelvic lymph nodes. However, the FN rate in PA lymph nodes detection decreases dramatically in case of negative PET/CT in pelvic region and surgical staging could be discussed in this case. Prospective trials are mandatory to validate this approach.