Impact of tumor histology on detection of pelvic and para-aortic nodal metastasis with FDG-PET in stage IB cervical cancer

Abstract

Objective:

FDG-PET detection of metastatic nodal disease is useful for guiding cervical cancer treatment, but the impact of tumor histology is unknown. This study reports the detection of FDG-avid pelvic and para-aortic lymph nodes in early stage cervical cancer patients with squamous carcinoma and adenocarcinoma tumor histology.

Methods:

Patients with FIGO 2009 stage IB1–2 cervical cancer who underwent pre-surgical FDG-PET between March 1999-February 2018 were identified in a tertiary academic center database. All patients had radical hysterectomy with pelvic and para-aortic lymph node dissection. The detection of pelvic and para-aortic lymph nodes by FDG-PET vs. surgical dissection was compared. FDG-PET sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were determined and stratified by tumor histology.

Results:

We identified 212 patients with early-stage cervical cancer (84% FIGO IB1, 16% IB2) who underwent pre-surgical FDG-PET; 137(65%) patients had squamous carcinoma and 75 (35%) patients had adenocarcinoma. PET/CT was performed in 189 (89%) patients and 23 (11%) had PET only. Surgical dissection revealed positive pelvic and para-aortic lymph nodes in 25% and 3.3% of patients, respectively. For squamous carcinoma, the sensitivity, specificity, PPV and NPV of FDG-PET for pelvic nodal metastasis were 44%, 99%, 95% and 78%; respectively. For adenocarcinoma, the corresponding results for pelvic nodal metastasis were 25%, 99%, 67% and 92%; respectively. The overall sensitivity, specificity, PPV and NPV of FDG-PET for para-aortic nodal metastasis was 29%, 99%, 67%, and 98%; respectively.

Discussion:

Pelvic nodal metastasis was less likely to be detected by FDG-PET in patients with early-stage adenocarcinoma than with squamous carcinoma.

Introduction:

In 2018, there were an estimated 569,847 new cases of cervical cancer worldwide accounting for 311,365 deaths ¹. The utility of ¹⁸F-fluorodeoxyglucose positron emission tomography (FDG-PET) has been well-established in staging and treatment planning of locally advanced cervical cancer ². Nodal metastatic disease is the most significant prognostic factor in predicting cancer recurrence and patient survival ³, leading to the recent change to allow advanced imaging such as FDG-PET in the 2018 International Federation of Gynecology and Obstetrics (FIGO) staging system ⁴.

The performance of FDG-PET for detecting lymph node metastasis in cervical cancer has been largely studied without attention to the impact of tumor histology on performance metrics. The prospective ACRIN 6671/GOG 0233 trial examined FDG-PET sensitivity and specificity for detecting abdominal/pelvic nodal metastasis in locally advanced cervical cancer patients, but only two patients had adenocarcinoma histology ⁸. For lung and gastric tumors, adenocarcinomas are known to be generally less FDG avid than squamous carcinomas ^9–11. Mucinous tumors are even less likely to be FDG avid ¹², and previous work from our group has found lower maximum standardized uptake values (SUVs) in non-squamous cervical primary tumors compared to squamous tumors ¹³. Therefore, we hypothesized lymph node detection by FDG-PET might be lower in cervical cancer patients with adenocarcinoma histology undergoing definitive surgical therapy.

Methods:

Women with cervical cancer who underwent pre-surgical FDG-PET between March 1999 and February 2018 were identified in a prospectively maintained patient database from an academic hospital that receives regional referrals for treatment of cervical cancer. Patients with FIGO 2009 stage IB cervical cancer who underwent PET and were subsequently referred for definitive chemoradiation (n = 161) were not included in this study. A total of 246 patients had pre-surgical FDG-PET prior to radical hysterectomy with pelvic and para-aortic lymph node dissection. Analysis of the whole cohort is displayed in supplemental Tables 1–3. For the purposes of this study, we focused on 212 patients with stage IB1–2 tumors (excluded IIA, n=4, and IIB, n=7) with either squamous or adenocarcinoma histology. Patients with adenosquamous (n=15), poorly differentiated (n=1), clear cell (n=3) and small cell (n=4) histologies were excluded. A subset of patients (n = 53) were previously reported on by Wright and colleagues in 2005 and are included in this current study ¹⁴.

For the work-up of early-stage cervical cancer, all patients had a pelvic and rectovaginal examination, routine laboratory testing and chest radiography; FDG-PET was ordered at the discretion of the attending gynecologic oncologist. Patients were clinically staged according to the FIGO 2009 system. From 1997–2002, FDG-PET was performed on a conventional PET scanner. Thereafter, all patients were imaged on hybrid FDG-PET/CT scanners. Patients were imaged approximately 60 minutes after 10–20 mCi of FDG (dosage dependent on body weight) was given intravenously. From 2004–2015, Foley catheterization and furosemide administration were routinely performed to improve discrimination of the bladder activity from the primary cervical tumor; its usage would not be expected to change the detection of pelvic and para-aortic lymph nodes. Except for the change in scanners over time and Foley catheterization, there has been no change in the patient preparation and image acquisition parameters, which have been previously described ¹⁴. FDG-PET scans were interpreted in a standard clinical manner.

Patients underwent either a Type II modified radical hysterectomy or Type III radical hysterectomy at the discretion of the surgeon. The pelvic lymph nodes were dissected around the external and internal iliac vessels as well as around the obturator fossa superior to the obturator nerve. The para-aortic nodes were removed from the bifurcation of the common iliac artery to the inferior mesenteric artery. Patients who received only a sentinel lymph node biopsy or radical hysterectomy without lymph node dissection were excluded. If a suspicious node was sent for frozen histological examination and was positive for disease, the procedure was usually aborted and the patient was sent for definitive chemoradiation. These patients were also excluded from this analysis. This retrospective study was approved by the institutional review board with waiver of informed consent (IRB# 201902127).

The surgically resected and imaged lymph nodes were categorized as pelvic if found in the common iliac, internal/external iliac, obturator, or pre-sacral nodal regions. The para-aortic region was defined as the nodal basins above the bifurcation of the aorta to the level of the inferior mesenteric artery. The pathologic results were considered the reference standard for pelvic and/or para-aortic nodal detection. Assessment of pelvic and para-aortic nodal region positivity was determined from the clinical radiology report on a per patient basis. The Fisher exact test was used to compare categorical data, as well as sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of FDG-PET detection of nodal metastases for squamous and adenocarcinoma tumor histology, respectively. Kaplan-Meier and Cox proportional hazards analysis were performed to compare freedom from relapse by histologic tumor type. Kaplan-Meier curves were compared using the log rank test. A p-value < 0.05 was considered statistically significant. Statistical analyses were done in SPSS, version 23 (IBM, Armonk, NY).

Results:

A total of 212 women with early stage cervical cancer including 137 (65%) with squamous carcinomas and 75 (35%) with adenocarcinomas were identified (Table 1). FDG-PET and FDG-PET/CT were performed in 23 (11%) and 189 (89%) patients, respectively. PET-positive pelvic and para-aortic lymph nodes were reported in 11% and 1.4% of patients; respectively. Pathologically positive pelvic and para-aortic lymph nodes were found in 25% and 3.3% of patients; respectively. Pelvic lymph nodes were pathologically detected in 33% of patients with squamous carcinoma histology compared to 11% with adenocarcinoma histology. The sensitivity of FDG-PET for detecting pelvic lymph nodes was higher for squamous carcinoma compared to adenocarcinoma (44% vs. 25%), and there was no difference in specificity (99% for both). Tables 2 and 3 summarize the frequency, sensitivity, specificity, PPV and NPV of FDG-PET for detecting pelvic and para-aortic lymph nodes, respectively. Supplemental Tables 4–9 show 2x2 tables comparing FDG-PET and surgically evaluated node positivity stratified by histology and nodal region.

Table 1:

Tumor stage and imaging stratified by histology.

Variables	Total No. of patients (%)	Squamous No. of patients (%)	Adenocarcinoma No. of patients (%)	P-value
Median Age (range)	44 (23–83)	44 (23–83)	41 (26–67)	0.05
Initial FIGO Stage				0.43
IB1	178 (84%)	113 (82%)	65 (87%)
IB2	34 (16%)	24 (18%)	10 (13%)
PET	23 (11%)	21 (15%)	2 (3%)	0.005
PET/CT	189 (89%)	116 (85%)	73 (97%)

p-value compares squamous vs. adenocarcinoma histology.

Table 2:

FDG-PET and pathologic detection of pelvic lymph nodes.

Variables	Total No. of patients (%)	Squamous No. of patients (%)	Adenocarcinoma No. of patients (%)	P-value
Frequency
FDG-PET	24/212 (11%)	21/137 (15%)	3/75 (4%)	0.01
Surgery	53/212 (25%)	45/137 (33%)	8/75 (11%)	<0.001
Sensitivity [95% CI]	41.5% [28–55]	44.4% [30–59]	25.0% [0–55]	0.45
Specificity [95% CI]	98.7% [91–100]	98.9% [97–100]	98.5% [96–100]	1.00
PPV [95% CI]	91.7% [81–100]	95.2% [86–100]	66.7% [13–100]	0.24
NPV [95% CI]	83.5% [78–89]	78.4% [71–86]	91.7% [85–98]	0.02

P-value compares squamous vs. adenocarcinoma histology.

Table 3:

FDG-PET and pathologic detection of para-aortic lymph nodes.

Variables	Total No. of patients (%)	Squamous No. of patients (%)	Adenocarcinoma No. of patients (%)	P-value
Frequency
FDG-PET	3/212 (1.4%)	2/137 (1.5%)	1/75 (1.3%)	0.94
Surgery	7/212 (3.3%)	6/137 (4.4%)	1/75 (1.3%)	0.24
Sensitivity [95% CI]	28.5% [0–62]	33.3% [0–71]	0% [0–0]*	1.00
Specificity [95% CI]	99.5% [99–100]	100% [100–100]	98.6% [96–100]	0.36
PPV [95% CI]	66.7% [13–100]	100% [100–100]	0% [0–0]*	0.33
NPV [95% CI]	97.6% [96–100]	97.0% [94–100]	98.6% [96–100]	0.66

• Only one patient with cervical adenocarcinoma histology had pathologically detected para-aortic node metastasis, which was not detected by FDG-PET.

• P-value compares squamous vs. adenocarcinoma histology.

Discussion

This study demonstrated different frequencies of FDG-PET detection of lymph node metastasis in patients with cervical adenocarcinoma vs. those with squamous carcinoma. Overall, the sensitivity for detecting nodal metastasis was under 50% in patients with clinical stage IB1 and IB2 disease who had been selected to undergo surgery. The sensitivity for detecting pelvic nodes in adenocarcinoma was nearly half that for squamous carcinoma. This is consistent with our prior finding that non-squamous primary cervical tumors are less FDG avid than are squamous carcinomas ¹³. The differential uptake of FDG in squamous carcinoma vs. adenocarcinoma may be related to tumor differentiation, as well as the tumor microenvironment. In general, fast-growing tumors with high cellularity, such as squamous cell carcinomas, have higher glucose metabolism than mucinous adenocarcinomas ^12,15. Inflammatory cells drawn to the tumor microenvironment have also been associated with increased FDG uptake ^18,19.

The overall sensitivity and specificity of FDG-PET in this cohort of patients with early-stage disease was comparable to those we previously reported in a smaller study ¹⁴ and to those found by Nogami and colleagues ²⁰. Other groups have reported higher sensitivity in detecting nodal metastasis in early-stage cervical cancer ^21,22. These retrospective study results are summarized in Table 4. None of these other studies stratified results by tumor histology, and differences in patient populations may partially explain the discrepancy in reported detection of nodal metastases with FDG-PET. The prospective ACRIN 6671/GOG 0233 trial reported higher overall sensitivity (81%) and lower specificity (69%) of FDG-PET for detecting abdominal/pelvis nodal metastasis in 153 locally advanced cervical cancer patients, but only 25 patients with stage IB2 would have been included in our early-stage study ⁸. Similar to our findings, the sensitivity for detecting abdominal nodes (50%) was less than that for detecting pelvic nodes (83%).

Table 4:

FDG-PET metrics for detecting pelvic lymph nodes in early stage cervical cancer

Variables	Sensitivity	Specificity	PPV	NPV
This study (n=212)
Overall	42%	99%	92%	84%
Squamous carcinoma	44%	99%	95%	78%
Adenocarcinoma	25%	99%	67%	92%
Nogami et al (n=70)	33%	93%	56%	84%
Dong et al (n=63)	88%	78%	39%	98%
Lv et al (n=87)	100%	Not reported	87%	100%

The frequency of histologically positive para-aortic nodes was only 3.3% in this cohort and did not significantly differ by histology. This is consistent with a prior prospective study from Korea, showing that the frequency of para-aortic nodal metastasis was 0% in early-stage cancer and increased to 8% with larger, more invasive primary tumors ²³. The reported sensitivity of FDG-PET in detecting para-aortic nodes in this study is statistically limited by the rarity of para-aortic metastasis in early stage cervical cancer. Comparisons between histological subtype were also impractical. However, our study suggests that in stage IB1-IB2 disease, the frequency of para-aortic nodal metastasis is low and surgical dissection of the para-aortic nodes after a negative FDG-PET may not be necessary.

The limitations of this single-institution study include its retrospective nature, which prevented standardized assessment of the lymph nodes. The number of nodes and regions sampled by the surgeons were not categorized for this analysis. The study does not definitively test the PET metrics for early stage cervical cancer because of the patient-level analysis rather than nodal-level analysis. Selection biases of the individual referring physicians meant not all stage IB cervical cancer patients treated at our institution received PET. While some patients with PET-positive nodes were triaged to definitive chemoradiation, the ones in this study may have had less obvious nodal involvement by imaging and received primary surgical treatment. This would lead to under-estimating the sensitivity of PET. The frequency of nodal metastasis in this selected population of patients with early-stage cervical cancer was relatively low, which can underestimate the PPV of FDG-PET, as seen previously ²⁴. However, the primary comparison of PET metrics between squamous carcinoma and adenocarcinoma histology should be equally affected by these limitations.

In conclusion, our study demonstrated that the frequency of positive pelvic nodes by histology was three-fold higher in squamous carcinoma than adenocarcinoma. FDG-PET was less sensitive detecting nodal metastasis in patients with cervical adenocarcinoma compared to those with the more common squamous histology.