Analysis of prognostic factors and clinical outcomes in uterine cervical carcinoma with isolated para-aortic lymph node recurrence

Abstract

To evaluate clinical outcomes and to identify prognostic factors in isolated para-aortic lymph node (PALN) recurrence, we retrospectively reviewed 65 patients who developed PALN recurrence as the first site of tumor progression from a total of 1521 patients who were treated with curative pelvic radiation therapy (RT) for uterine cervical carcinoma between May 1993 and January 2017. Forty-five of the 65 patients received salvage therapy. The median salvage PALN radiation dose was 54 Gy (range: 18 to 62 Gy). Prognostic factors for overall survival (OS) and distant metastases (DMs) were analyzed with univariate and multivariate Cox regression. The median follow-up period for surviving patients was 61 months (4-202 months). The median OS was 27.7 months (0.3-202 months). The highest level of PALN metastases at or above the L1 spinal level (hazard ratio [HR] 9.88, 95% confidence interval [CI] 3.44-28.38, P<0.001) and the presence of leg edema and/or back pain at recurrence (HR 3.25, 95% CI 1.57-6.75, P=0.002) were significantly associated with worse OS. A significantly higher incidence of DMs (HR 5.97, 95% CI 2.05-17.35, P=0.001) was found in the patients with a high level (≥L1) of PALN metastases. Salvage RT (HR 0.35, 95% CI 0.17-0.71, P=0.004) and restaging with positron emission tomography/computed tomography (PET/CT) (HR 0.2, 95% CI 0.04-0.93, P=0.039) were independent predictors of a better OS. In conclusion, a high level (≥L1) of PALN metastases predicts poor survival and a high rate of DMs. Periodic surveillance for early detection and restaging by PET/CT imaging to identify the optimal treatment at recurrence is recommended.

Introduction

Pelvic irradiation is one of the mainstay treatment modalities for cervical cancer without para-aortic lymph node (PALN) metastases in either an adjuvant [1,2] or definitive setting [3-6]. Compared with lower pelvic recurrence in the initial treatment field, metastatic sites beyond the pelvis, including the PALN, represent the largest proportion of treatment failure [7,8]. PALNs are a sanctuary for microscopic disease and are a common failure site [9]. Berman et al. [10] reported a progressive rise in the incidence of PALN metastases with increasing clinical stage, ranging from 5 to 25% in stage I-III patients. Additionally, Huang et al. [7] reported that three crucial risk factors (high squamous cell carcinoma antigen (SCC-Ag), positive pelvic lymph nodes and advanced parametrial invasion) predicted a greater risk of PALN recurrence after primary radiotherapy. Once PALN failure occurred after primary pelvic irradiation, the prognosis was poor, with a 5-year survival rate of less than 30% [7]. However, for patients with PALN relapse alone, Hong et al. [8] suggested that curative treatment is possible for whom treatment has a palliative intent, in contrast to the treatment of patients with supraclavicular lymph nodes (SCLNs) with or without PALN involvement or at sites other than the PALNs and SCLNs. In addition, a previous study [11-13] showed durable salvage treatment outcomes for patients with isolated PALN relapse who received concurrent chemoradiation. Therefore, the identification of prognosticators of survival is required to precisely estimate the treatment effect. In addition, there have been no published reports on the role of the characteristics of the lymph node status, such as the highest level of PALN metastases and the use of positron emission tomography (PET) in isolated PALN recurrence. The aim of this study was to review the outcomes of patients with isolated para-aortic recurrence after primary or postoperative radiation therapy (RT) and to investigate the independent prognostic factors, including imaging information, for overall survival (OS).

Materials and methods

Patient characteristics

From May 1993 to January 2017, 1521 patients with Stage IA to IVA primary uterine cervical carcinoma according to the International Federation of Gynecology and Obstetrics (FIGO) classification were initially treated with curative pelvic irradiation in either the definitive or postoperative setting. Of these patients, 65 developed PALN recurrence as the first site of tumor progression and were retrospectively reviewed. The exclusion criteria included PALN metastases confirmed by histology or suspected by imaging prior to the initiation of primary RT or during primary RT, treatment with prophylactic extended-field RT, or cervical stump cancer after total hysterectomy. Our protocols were reviewed and approved by the Institutional Review Board (IRB) of the Chang Gung Medical Foundation (IRB No. 201901065B0).

Initial radiation treatment

All 65 patients underwent a combination of external-beam RT (EBRT) with or without systemic chemotherapy and intracavitary brachytherapy; these therapies were considered definitive or adjuvant therapy. Typically, the treatment method involved 39.6-45 Gy (1.8-2 Gy per fraction) of whole-pelvis irradiation with 10-15 megavoltage (MV)-X ray by anteroposterior parallel-opposed, four-field box radiation or intensity-modulated RT (IMRT). A low-pelvic boost (range: up to 50.4-59.4 Gy) was administered with or without a 4-cm midline central block for patients with locally advanced disease. Para-aortic region irradiation was not performed. After external-beam radiation in the definitive setting, the Henschke applicator was used for intracavitary brachytherapy with an Ir-192 high-dose-rate unit (MicroSelectron, Nucletron, Veenendaal, the Netherlands) and an afterloading technique. A total of 24 Gy in five fractions (1993-2001) and either 24 Gy in four fractions or 27 Gy in six fractions (2001-2017) was prescribed as a typical RT dose to point A. In addition, a 2D-based planning system with radiographic film was used to evaluate the dose to the bladder and rectum points of the International Commission on Radiation Units and Measurements, report #38. However, because the applicator for brachytherapy could not be inserted into the cervical canal, 2 patients received an IMRT boost up to 66.6 Gy and 68.4 Gy, respectively, instead of brachytherapy at the end of pelvic irradiation. Among the patients who initially underwent concurrent chemoradiation, the systemic regimen consisted of cisplatin (50-75 mg/m²) and 5-fluorouracil (500-1000 mg/d/m² for 4 days) or weekly cisplatin (30 mg/m²).

Posttreatment follow-up

After the initial treatment with EBRT and brachytherapy, posttreatment surveillance was performed along with physical examinations according to a follow-up schedule that involved visits every month for the first 3 months, every 3-4 months for the first 2 years, and every 6-12 months thereafter. Biologic investigations during follow-up involved tumor markers (carcinoembryonic antigen (CEA) and SCC) every 1-2 months after the completion of treatment and every 3-6 months thereafter or whenever any suspicious clinical symptoms were present. Chest radiography and pelvic-abdominal computed tomography (CT) were also performed after radiation treatment and then annually or whenever there was suspicion of recurrence based on clinical symptoms or elevated tumor markers.

Isolated para-aortic recurrence and salvage treatment

All patients had a disease-free period of more than 1 month, with either the disappearance of the cervical tumor and parametrial induration or the normalization of tumor markers after primary radiation during follow-up, confirming that there was not a persistent locoregional residual tumor that had metastasized to the PALN. We defined isolated para-aortic recurrence as an enlarged PALN (>10 mm diameter of the short axis) by imaging studies with neither simultaneous locoregional recurrence nor distant relapse. In addition, PET/CT scanning was used to exclude the possibility of other distant metastatic sites at the discretion of the treating physicians. Among the 65 patients who developed isolated PALN recurrence, the highest level of PALN metastases on CT images was measured and recorded with parallel levels of the vertebral segment. In addition, if the highest PALN was located parallel to a disc between vertebral segments, the highest level of the PALN was recorded as the upper level of the vertebral segment. The salvage treatment modality was determined by the consensus of both gynecologic oncology and radiation oncology physicians. Typically, the salvage radiation dose given to the PALN region was 39.6-45 Gy with or without a gross-tumor boost up to 54-62 Gy, with a fraction size of 1.8-2.2 Gy per day by 10 MV photons. Gross PALN irradiation alone with 54-60 Gy was an alternative treatment.

Statistical analysis

Survival after recurrence was defined as the length of time from isolated PALN recurrence to the last follow-up visit or death and was calculated by the Kaplan-Meier method. The time to distant metastases (DMs) was defined as the length of time from PALN recurrence to the last follow-up of any kind or to imaging evidence or pathologic proof of DMs other than PALN. The statistical significance of differences among groups was compared using the log-rank test. A Cox proportional hazards model was used to perform multivariate analysis with stepwise forward selection. Five variables were input for regression models. Symptoms including leg edema or back pain, whether salvage RT was received and whether 18F-fluorodeoxyglucose (FDG) PET/CT was used to assess isolated PALN recurrence were evaluated as binary data. Parameters such as recurrent tumor markers (SCC-Ag and CEA) and the highest level of PALN metastases were regarded as categorized data with two cutoff levels and were included in the multivariate analyses. Statistical SPSS software version 18.0 was used for all data processing.

Results

Characteristics of the patients with PALN recurrence

The median duration from the end of the initial treatment to isolated para-aortic recurrence was 21 months (1-226 months). The median follow-up for surviving patients with isolated PALN recurrence was 61 months (4-202 months). Patient characteristics are listed in Table 1. The median age was 58 years (36-84 years). Sixty-two (95.3%) patients had SCC based on histopathology.

Table 1.

Patient characteristics

Parameter	Number (%)
Age (years)
<45	9 (13.8%)
45-70	44 (67.7%)
>70	12 (18.5%)
Histopathology
SCC	62 (95.3%)
Adenocarcinoma	3 (4.6%)
Clinical stage at the initial treatment
Ib	15 (23.1%)
IIa	4 (6.2%)
IIb	34 (52.3%)
IIIa	1 (1.5%)
IIIb	11 (16.9%)
IVa	0
SCC-Ag level at recurrence (ng/ml)
≥10	15 (23.1%)
<10	37 (56.9%)
Unknown	13 (20%)
CEA level at recurrence (ng/ml)
≥10	9 (13.8%)
<10	42 (64.6)
Unknown	14 (21.5%)
Symptoms
None	39 (60%)
Lumbago	19 (29.2%)
Edema of the lower extremities	10 (15.4%)
Mean PALN size (cm) ± SEM	1.78 ± 0.11
PET/CT restaging
Yes	11 (16.9%)
No	54 (83.1%)
Highest level of PALN metastases
T12-L1	28 (43.1%)
L2-L4	21 (32.3%)
Unknown	16 (24.6%)
Salvage treatment
Chemotherapy alone	10 (15.4%)
Radiotherapy alone	11 (16.9%)
Concurrent chemoradiation	25 (38.5%)
None	19 (29.2%)

SCC-Ag, squamous cell carcinoma antigen; CEA, carcinoembryonic antigen; PALN, para-aortic lymph node; SEM, standard error of the mean.

Only three (4.6%) patients had adenocarcinoma. Regarding the clinical stage at the initial treatment, more than half of the patients (n=34) were found to have FIGO stage IIb. The initial clues that led to screening for PALN recurrence are summarized in Table 2. Twenty-five (38.5%) of the 65 patients had elevated tumor markers only, and 10 (15.4%) patients had symptoms only as the first evidence of PALN recurrence.

Table 2.

Initial evidence of PALN recurrence

Initial presentation	N (%)
Elevated tumor marker combined with symptoms	17 (26.2%)
Elevated tumor marker only	25 (38.5%)
Symptoms only	10 (15.4%)
Routine positive finding of pelvic CT	13 (20%)

PALN, Para-aortic lymph node; CT, Computed tomography.

Regarding symptoms, 26 (40%) patients had symptoms associated with PALN recurrence, with lumbago as the most frequent symptom, which was observed in 19 patients. Edema of either the unilateral or bilateral lower extremities was also observed in 10 patients. The tumor markers SCC (except for in 13 patients) and CEA (except for in 14 patients) were detected at recurrence. Thirty of 52 (46.1%) patients had an elevated SCC level, with a median value of 9.65 ng/ml, ranging from 2.57 to 150 ng/ml. Additionally, nine patients (13.8%) had normal SCC-Ag levels but elevated CEA levels, ranging from 6.1 to 96.7 ng/ml (median: 9.7 ng/ml). Notably, 13 of 65 (20%) patients had neither symptoms nor elevated tumor markers during routine follow-up. Concerning lymph node metastasis characteristics, the mean size of the PALN was 1.78 ± 0.11 cm. Images were available for 49 of the 65 patients, and of these, the highest levels of PALN metastases were found at T12, L1, L2, L3 or L4 in eight (12.3%), 21 (32.3%), 10 (15.4%), seven (10.8%) and three (4.6%) patients, respectively. In addition, the proportion of patients with elevated tumor markers was 54.2% and 15.2% in patients with T12-L1 and L2-L4 level PALN metastases, respectively (P=0.007).

Salvage treatment and survival outcomes

Among these 65 patients, 10 patients underwent chemotherapy (CT) alone, 11 had para-aortic irradiation (RT) alone, and 25 underwent salvage concurrent chemoradiotherapy (CCRT), while 19 patients did not receive further treatment and received palliative care due to old age, poor performance status or severe urosepsis. The median RT dose was 54 Gy (18 Gy-62 Gy) in the patients undergoing salvage RT or CCRT. One patient received incomplete radiation treatment, with an RT dose of only 18 Gy, because the patient refused further RT due to general fatigue and intolerable diarrhea. The median OS was 27.7 months (0.3-202 months), and the 5-year survival rate after PALN recurrence was 33.9%. Stratified by salvage treatment, the 5-year OS rate for patients undergoing salvage CCRT, RT or CT was 48.4%, 44.7%, and 40%, respectively, and no significant difference was found (P=0.231). In patients who did not undergo salvage treatment, the 5-year OS was 0%. Among patients with the highest level of PALN metastases (L1 or above), a higher 5-year survival rate (31.8%) was observed in patients undergoing salvage RT (including RT alone or CCRT) than in patients undergoing salvage CT alone (0%) (P=0.003).

Univariate and multivariate analyses for OS

Table 4 shows the results of the univariate and multivariate analyses for OS. The highest level of PALN metastases at L1 or above (HR: 9.88, 95% CI: 3.44-28.38, P<0.001) and the presence of symptoms with recurrence (HR: 3.25, 95% CI: 1.57-6.75, P=0.002) were independent predictors of lower OS. In addition, salvage RT (HR: 0.35, 95% CI: 0.17-0.71, P=0.004) and evaluation with PET-CT (HR: 0.2, 95% CI: 0.04-0.93, P=0.039) showed a survival benefit in multivariate analysis. The 5-year OS for patients who underwent PET-CT to exclude other DMs at recurrence was 81.8%, and 90.9% (10/11) of those patients underwent salvage CCRT. In addition, based on restaging by PET/CT, eight patients were asymptomatic at recurrence and had a 5-year survival rate of 87.5% after salvage CCRT. In the analyses of patients treated with salvage RT or CCRT (salvage RT group), the corresponding 5-year OS rates for patients with PALN recurrence at ≥L1 or <L1 levels were 31.8% and 74.9%, respectively (P=0.013) (Figure 1A). In addition, patients in the salvage RT group who underwent PET/CT restaging showed a higher 5-year OS rate (81.8%) than those who did not undergo PET/CT restaging (33.9%) (P=0.041) (Figure 1B). For patients with or without symptoms in the salvage RT group, the corresponding 5-year OS rates were 26% and 58%, respectively (P=0.004) (Figure 1C).

Table 4.

Univariate and multivariate analyses of OS rates

Parameters	UVA		MVA
	5-y OS (%)	p-value	HR (95% CI)	P-value
Tumor marker: SCC-Ag ≥10 or CEA ≥10 (yes vs no)	21.9 vs 50.9	0.006	-	0.59
Symptoms: lumbago or leg edema (yes vs no)	18.6 vs 42.7	0.002	3.25 (1.57-6.75)	0.002
Highest level of PALN (at L1 or above vs below L1)	17.4 vs 74.3	<0.001	9.88 (3.44-28.38)	<0.001
Salvage RT (yes vs no)	47 vs 16.6	<0.001	0.35 (0.17-0.71)	0.004
PET/CT (yes vs no)	81.8 vs 25	0.005	0.2 (0.04-0.93)	0.039

UVA, univariate analysis; MVA, multivariate analysis; RT, radiation therapy; SCC-Ag, squamous cell carcinoma antigen; CEA, carcinoembryonic antigen; PET/CT, positron emission tomography/computed tomography; HR, hazard ratio; CI, confidence interval.

Figure 1.

Kaplan-Meier estimates of overall survival in patients undergoing salvage CCRT or RT. A. The 5-year OS rates for patients with levels of PALN recurrence ≥L1 and <L1 were 31.8% and 74.9%, respectively (P=0.013). B. The 5-year OS rates were 81.8% and 33.9% for patients with or without PET/CT restaging, respectively (P=0.041). C. For patients with or without symptoms, the corresponding 5-year OS rates were 26% and 58%, respectively (P=0.004).

Distant metastases followed by PALN recurrence

Subsequent DMs following isolated PALN recurrence were observed in 24 (36.9%) of 65 patients. Table 3 shows the site of secondary metastases in detail. The most frequent site of DMs was the supraclavicular lymph nodes, followed by the lung. The median time to DMs after PALN metastases was 49.6 months (1-202 months), and the actual 5-year incidence of DMs was 55.7%. Table 5 shows the results of the univariate and multivariate analyses for DMs. The highest level of PALN metastases at or above L1 (HR: 5.97, 95% CI: 2.05-17.35, P=0.001) was an independent predictor of lower OS. In patients undergoing salvage CCRT or RT, the 5-year cumulative DM incidences were 29.4% and 78.5%, respectively, for PALN recurrence at or below L1 (P=0.01) (Figure 2).

Table 3.

Site of secondary metastases outside of PALN after PALN recurrence

Treatment/secondary recurrence site	CCRT	RT	CT	Palliative care	Total
Bone	2	2	1	2	7
Peritoneal	1	0	0	0	1
SCLN	5	3	0	4	12
Mediastinal LN	3	2	1	2	8
Lung	5	3	0	2	10
Pleural	0	0	1	0	1
Liver	1	1	1	0	3
Inguinal LN	1	0	0	0	1

CCRT, concurrent chemoradiation; RT, radiation therapy; CT, chemotherapy; LN, lymph node; SCLN, supraclavicular lymph node.

Table 5.

Univariate and multivariate analyses of DM rates

Parameters	UVA		MVA
	5-y DM (%)	p-value	HR (95% CI)	p-value
Tumor marker: SCC-Ag ≥10 or CEA ≥10 (yes vs no)	48 vs 55	0.934	-	0.325
Symptoms: lumbago or leg edema (yes vs no)	54.6 vs 54	0.25	-	0.393
Highest level of PALN (at L1 or above vs below L1)	83.7 vs 28.9	<0.001	5.97 (2.05-17.35)	0.001
Salvage RT (yes vs no)	57 vs 49.9	0.844	-	0.837
PET/CT (yes vs no)	45.5 vs 58.8	0.441	-	0.895

UVA, univariate analysis; MVA, multivariate analysis; RT, radiation therapy; SCC-Ag, squamous cell carcinoma antigen; CEA, carcinoembryonic antigen; PET/CT, positron emission tomography/computed tomography; HR, hazard ratio; CI, confidence interval.

Figure 2.

Prognostic comparison of the patients undergoing salvage CCRT or RT in distant metastasis curves, stratified by the highest level of PALN metastases. For patients with levels of PALN recurrence ≥L1 and <L1, the 5-year cumulative DM incidences were 78.5% and 29.4%, respectively (P=0.01).

Discussion

After curative pelvic RT for invasive uterine cervical carcinoma, the incidence of isolated PALN recurrence ranged from 1.6% to 12% [7,8,11,13-15]. Several predictors of OS were identified in these patients in a previous study, which included tumor markers (SCC-Ag) [11], the presence of symptoms after recurrence [11,13,16], the salvage method [11], and the time to recurrence [12,13]. In the present study, a similar incidence of isolated PALN recurrence (4.3%) was observed. We found that the highest level of PALN metastases, the presence of symptoms, salvage RT and use of PET-CT scanning for restaging were important prognostic factors for OS.

The outcome of isolated PALN recurrence was still poor. Grigsby et al. [15] reported a median survival of only 8.7 months after isolated PALN recurrence and a 100% mortality rate within two years of para-aortic recurrence. In another study, the 5-year survival rate of patients with isolated relapse ranged from 15.8% to 45% [7,8,11,13,16]. The survival rate (36%) in our study was comparable to that in a previous report. The standard treatment of isolated PALN recurrence in cervical cancer is not well established. A series reported by Chou et al. [11] showed that asymptomatic patients with isolated PALN recurrence who received concurrent chemoradiation as salvage treatment had better long-term survival, with a 5-year survival rate of 50%. Prognostic factors and survival outcomes for isolated PALN recurrence are listed in Table 6. It should be noted that Singh et al. [13] reported a five-year survival rate of 100% for seven patients who received full-dose salvage concurrent chemoradiation. In contrast, Kim et al. [12] reported a 3-year survival rate of only 19% for 12 patients with isolated PALN recurrence after treatment with hyperfractionated RT and concurrent chemotherapy. By examining the characteristics of the seven patients in the Singh et al. study [13], we found that none exhibited classical symptoms at recurrence, and they were all diagnosed by PET scan. Thus, excellent survival outcomes are found to correspond to good prognosticators, including receiving salvage RT, PET-CT scanning for restaging, and no symptoms at recurrence, as indicated in the present study. In addition, asymptomatic patients who were diagnosed by PET/CT and who underwent salvage CCRT in the current study had similar outcomes, with a 5-year survival rate of 87.5%.

Table 6.

Literature review of outcomes for isolated PALN metastases

Investigator	Symptoms at recurrence	Use of PET or PET/CT	Salvage CCRT or RT	High level of PALN metastases (≥L1)*	RT dose (Gy)	Outcome
Grigsby et al. [15] 1994 (n=20)	85%	0%	RT: 100%	N/A	46.4 (median)	Median OS: 8.7 months
Chou et al. [11] 2001 (n=26)	46.2%	0%	RT: 3.8%	N/A	45	5-y OS: 30.8%
			CCRT: 53.8%			(CCRT-5-y OS: 50%)
Kim et al. [12] 2003 (n=12)	33%	0%	CCRT: 100%	41.6% (5/12)	60 (1.2 Gy/1 fr)	3-y OS: 19%
Hong et al. [8] 2004 (n=46)	50%	0%	RT: 21.7%	N/A	40-50	5-y OS: 27%
			CCRT: 54.3%			(CCRT-5-y OS: 40%)
Singh et al. [13] 2005 (n=14)	50%	42.9%	RT: 14.2%	N/A	50.4 (median)	5-y OS: 50%
			CCRT: 50%			(CCRT-5-y OS: 100%)
Niibe et al. [16] 2006 (n=84)	21.4%	N/A	RT: 61.9%	N/A	50.8 (mean)	5-y OS: 31.3%
			CCRT: 38.1%			(CCRT-3-y OS: 37.7%)
Present study (n=65)	44.6%	18.5%	RT: 16.9%	43.1% (28/65)	54 (median)	5-y OS: 36%
			CCRT: 38.5%			(CCRT-5-y OS: 52%)

N/A, not applicable; OS, overall survival; fr, fraction; CCRT, concurrent chemoradiotherapy; RT, radiation therapy; PET/CT, positron emission tomography/computed tomography.

^*The highest level of PALN over the disc between L1 and L2 was recorded as L1 in the present study.

Although patients undergoing salvage RT showed a survival benefit, there was no significant difference among patient outcomes for those who received salvage RT, salvage CT or salvage CCRT when excluding patients who underwent palliative care. However, for patients with the highest level of PALN at L1, the addition of local RT (CCRT or RT alone) seemed to play an important role, with a higher OS (31.8%) observed than that with CT alone (0%). In addition, a previous retrospective study showed no significant difference in survival between salvage RT and salvage CCRT [8,16]. Local salvage treatment, including CCRT or RT alone, may be a feasible treatment option, but the optimal salvage method requires a prospective study for validation.

The radiologic feature of the highest level of PALN metastases has been analyzed as a prognostic factor for survival in a few retrospective studies [17,18]. Jang et al. [17] found no correlation between survival and the level of PALN when classifying PALN metastases of the lumbar vertebra as ≥ level 3 and ≤ level 4, but the results of those classified as ≥ level 1 and ≤ level 2 were not reported. Wu et al. [18] demonstrated that there was no significant difference in survival between patients with differences in the highest levels of PALN metastases (L1-L2 vs L3-L4, P=0.133). However, the previous study mentioned above [17,18] was performed in the setting of patients with cervical cancer with initial PALN metastases, where the characteristics of the primary cervical tumor also may have served as major prognostic factors and might have masked the impact of the location of the PALN on survival. To the best of our knowledge, no data regarding the association between survival and the highest level of PALN metastases were available in the cohort with isolated PALN recurrence after pelvic irradiation. In the current study, the highest level of metastases was found to be an important prognostic factor. Lymph node metastases occurred with a stepwise progression for cervical cancer [10,19], and PALN metastases were found to be the midway point between local and systemic disease. Based on our results, a higher level of PALN metastasis (≥L1) was associated with a higher probability of DMs, which decreased survival. The possible mechanistic basis is best understood through its lymphatic drainage and the concepts of metastatic cascade and sequential metastases in which tumors seed lymph node metastases and then further seed distant metastases [20,21]. Para-aortic lymph nodes have been categorized as suprarenal and infrarenal subgroups demarcated by the renal vein, for which the most common spinal level is between the first and second lumbar vertebrae [22]. The distribution of the suprarenal and infrarenal para-aortic lymph node regions corresponded with spinal levels ≥L1 and <L1, respectively. To elucidate the regional spread in cervical cancer, Höckel et al. [23] demonstrated the topography and stepwise pattern of para-aortic lymph node metastases from the inframesenteric, infrarenal to suprarenal nodes in order, after which metastases integrated into the lymph stream. Through an analysis of nodal failure in that study, the metastatic tumor cells from the lower level (<L1) of the para-aortic lymph node, including the inferior mesenteric lymph node and infrarenal area, travelled downstream most frequently to the closest nodes in the lymph basin and the suprarenal nodes that were still confined to the regional lumbar compartment [23]. In addition, the thoracic duct started at the T12-L1 level initially, with the cisterna chyli receiving the lumbar trunks formed by the union of the efferent vessels from para-aortic lymph nodes [24-26]. The higher PALN level (≥L1) seems to be the closest location among the lumbar trunks to the thoracic duct, which communicates with the systemic venous circulation at the junction of the left subclavian and left internal jugular vein as a side loop of the hematogenous metastatic circuit [24,27]. In addition, the left-sided supraclavicular lymph nodes represent the final common pathway for the drainage of infradiaphragmatic lymph nodes [28] and were the most common distant metastatic site found in the current study. Therefore, based on the outcomes in the current study, isolated PALN recurrence is regarded as locoregional disease at a lower level (<L1) but systemic disease at higher levels (≥L1). More aggressive treatment modalities, such as CCRT, might be considered for patients with PALN metastases at high levels.

CT or MRI scans are common tools for the evaluation of recurrent cervical cancer. However, both imaging techniques principally detect lesions based on the morphological change and size and have a limited scanning range. Therefore, enlarged reactive lymph nodes cannot be differentiated from metastatic nodes, and tiny lymph nodes may be missed. PET/CT provides both metabolic and anatomic information and more accurately detects lesions in a recurrent setting [29-31]. Mittra et al. [30] reported a sensitivity of 93-96% and a specificity of 93-95% for PET/CT for the detection of local and distant recurrence. In addition, Lai et al. [32] found that in patients who underwent salvage surgery for recurrence, a significantly better 2-year survival (HR: 0.21, 95% CI: 0.05-0.83, P<0.020) was observed among the cohort who underwent disease restaging with PET than that for patients who did not undergo PET. In the present study, the use of PET/CT for restaging was associated with improved survival in patients who underwent salvage RT or CCRT (P=0.048). In accordance with the superiority of PET/CT over CT or conventional imaging for the detection of metastatic lesions [33,34], the confirmation of isolated PALN recurrence by PET/CT can more accurately exclude the possibility of occult metastases beyond PALN or the local recurrence of a primary tumor that already exists and could potentially decrease survival rate. In addition, the radiation treatment volume can be modified to specifically target the FDG-positive tumor based on PET/CT images. Bjurberg et al. [33] reported that FDG-PET findings led to alterations in the treatment dose, the planned procedure or the mode of delivery during the management of recurrent cervical cancer in 22% of patients. Esthappan et al. [35] used PET/CT to effectively guide IMRT, which enabled a higher dose boost to positive PALNs and spared adjacent normal tissue. In a cohort of patients with isolated PALN recurrence, Singh et al. [13], as mentioned above, detected asymptomatic isolated PALN recurrence by CT, and PET appeared to allow durable survival with salvage CCRT (Table 6). Therefore, restaging with PET/CT is recommended for guiding the treatment plan of patients with isolated PALN recurrence.

Both elevated pretreatment SCC-Ag [36-40] and elevated CEA with normal SCC-Ag levels [36,39] were independent predictors for OS and recurrence in cervical cancer patients undergoing radiotherapy. However, only a few studies have discussed the role of tumor markers in the OS of patients with isolated PALN recurrence [7,11,12,16], with low sample sizes and univariate analysis. Chou et al. [11] reported that patients with SCC-Ag levels higher than 4.0 ng/ml had poor survival, with only one patient surviving among 16 patients. Huang et al. [7] showed a lower 5-year survival rate in patients with an SCC-Ag level of ≥10 ng/ml than in patients with levels <10 ng/ml (17% vs 66%) but only with a short follow-up (median, 31 months). Niibe et al. [16] reported a nonsignificant reduction in survival (3-year survival: 35.7%) for patients with an SCC-Ag level of ≥10 ng/ml compared with that in patients with levels <10 ng/ml. In the current study, CEA was also analyzed based on a previous study that reported a poor prognosis (including OS and DM) in patients with CEA levels ≥10 ng/ml and normal SCC-Ag levels [36,39] and a correlation between tumor progression (invasion and DMs) and the secretion of CEA in cervical cancer [41,42]. Tumor markers were not found to be independent predictors of OS in the multivariate analysis of the current study. The key predictor of OS based on these results was the highest level of PALN metastasis, which indicates the extent of LN spread and is reflected by elevated tumor markers, according to the high correlation between elevated tumor markers and the highest level of PALN metastases in the present study. More patients with T12-L1 metastases than patients with L2-L4 metastases had elevated tumor markers (either CEA or SCC) (54.2% vs 15%, P=0.007). Therefore, imaging findings of LN spread might provide important information for predicting OS that is more accurate than tumor markers. Additionally, 64.7% of patients had elevated tumor markers as the initial clues for detecting isolated recurrence. Therefore, periodic imaging surveillance was required to complement the other early detection methods and to provide prognostic information.

There are some limitations in the present study. Our study was retrospective and covered over 20 years, and heterogeneity among patient characteristics, chemotherapy regimens or RT techniques resulted in an inherent bias. In addition, the small number of patients precluded powering conclusive data, such as the optimal salvage RT dose or treatment modality. Additionally, all metastatic PALNs were clinically diagnosed by either CT or PET-CT imaging, without pathological confirmation by biopsy. Imaging studies were not available for a few patients because of the loss of film-based imaging.

Conclusions

In conclusion, the highest level of PALN metastases at the L1 spine level or higher and the presence of symptoms predicted worse OS in the isolated PALN recurrence of cervical cancer. RT plays an important role in improving OS. Periodic surveillance, including imaging studies, is recommended for early detection. Additionally, PET/CT served as a better restaging tool to exclude occult metastases and provide accurate information for the treatment of isolated PALN recurrence.

Disclosure of conflict of interest

None.