Objective
The purpose of this study was to explore in greater depth the outcomes of the Italian randomized trial investigating the role of pelvic lymphadenectomy in clinical early stage endometrial cancer. In the attempt to identify the patients with poorer prognosis, the impact of age and body mass index were also thoroughly investigated by cancer-specific survival (CSS) analyses.
Study Design
Survival outcomes of trial patients were analyzed in relation to age (≤65 years and >65 years) in the 2 arms (lymphadenectomy and no lymphadenectomy) and in the whole population of the trial.
Results
Univariate and multivariable analyses of CSS and overall survival (OS) of patients showed that age >65 years is a strong independent poor prognostic factor (5-y OS 92.1% and 78.4% in ≤65 years and >65 years patients, respectively, P < .0001; 5-y CSS 93.8% and 83.5% in ≤65 years and >65 years patients, respectively, P = .003). Among women ≤65 years, node negative patients had 94.4% 5-y OS and 96.3% 5-y CSS vs 74.3% 5-y OS and 74.3% 5-y CSS for node positive patients ( P = .009 and P = .002, respectively), while among women >65 y, node negative patients had 75.7% 5-y OS and 83.6% 5-y CSS vs 74.1% 5-y OS and 83.3% 5-y CSS for node positive patients ( P = .55 and P = .58, respectively). Univariate and multivariable survival analyses in the whole trial population showed that older age, and higher tumor grade and stage were significantly associated to a worse prognosis.
Conclusion
Older women faced an intrinsic poorer survival whether or not they underwent lymphadenectomy, and, unexpectedly, irrespective of the presence of nodal metastasis. Only in older patients was obesity (body mass index >30) significantly associated with scarce prognosis.
Endometrial cancer currently represents the most frequent gynecologic malignancy in western countries, and the seventh most common cancer in women. Incidence rates have been increasing during recent decades, particularly among postmenopausal women, along with the prolongation of life in the developed world.
Despite the generally good prognosis for the majority of patients, it is estimated that in 2012 endometrial carcinoma was the cause of more than 8000 deaths in the United States.
Some aspects of its natural history according to different histologic types and pathogenetic pathways are probably scarcely understood and, with their proper treatment, remain a topic of continuing controversy.
The Italian prospective randomized multicenter trial investigating the role of pelvic lymphadenectomy in early stage endometrial cancer began from some retrospective analyses showing apparent benefit in survival for patients submitted to pelvic lymphadenectomy in addition to standard total hysterectomy with bilateral adnexectomy.
The final results of the Italian trial on the role of systematic pelvic lymphadenectomy clearly failed to identify any positive impact of nodal resection on survival in clinical early stage endometrial cancer.
The above findings were also subsequently confirmed in the British ASTEC Trial (A study in the treatment of endometrial carcinoma), contributing to the later Cochrane metanalysis on this topic. The results have sparked heated scientific controversies, and these studies actually seem to have raised more questions and debating issues than they answer.
During recent years, new prognostic factors have come to light in literature and have been the object of study and discussion.
In particular, the prognostic role of age in endometrial carcinoma is debated. Although some authors identified older age as a strong independent poor prognostic factor, over the last decade some other studies have linked the worse survival outcomes in older patients to the presence of comorbidities, nonendometrioid histology tumors, and adjuvant treatments that would be less frequently performed in such patients.
To explore the outcomes of our trial in greater depth, and in the attempt to identify the patients with poorer prognosis, we further analyzed prognostic factors among patients from the study population. Specifically, the prognostic impact of age and body mass index (BMI) were also thoroughly investigated by cancer-specific survival (CSS) analyses.
Materials and Methods
All analyses were performed on women from the database of patients enrolled in the Italian multicenter randomized trial evaluating the role of pelvic systematic lymphadenectomy in supposed early stage endometrial carcinoma.
Women eligible for the trial were patients with histologically proven endometrioid or adenosquamous endometrial carcinoma clinically confined to uterus (preoperative International Federation of Gynecology and Obstetrics [FIGO] stage I disease). ( ClinicalTrials.gov no. NCT00482300 ). The study protocol was first accepted by local ethic committees, and proper written consent was required from all eligible patients.
Patients’ accrual, characteristics and follow up, as well as type of randomization, surgical procedures and adjuvant treatments performed are previously described in the original paper.
As clearly shown in the previously published analyses, age was significantly associated to survival outcomes in both univariate and multivariable analyses; in these analyses, age 65 years was identified by receiver operating characteristic curve as the optimal cutoff value for this prognostic factor.
Because pelvic lymphadenectomy was not significantly associated to survival outcomes, in these analyses instead of lymphadenectomy, we chose to use different prognostic factors (ie, age, BMI) for the whole trial population.
For the same reason, myometrial invasion was excluded from current analyses, because it was not significantly associated with overall survival (OS) in the original multivariable analysis.
BMI, on the contrary, was not included in the original univariate and multivariable model for progression free and OS, and its prognostic value and its interaction with patients’ age was thoroughly explored in the current analysis.
Statistical analyses
OS, the primary outcome of the trial, was defined as the time from random assignment to death from any cause. Disease-free survival was defined as the time from random assignment to the earliest occurrence of relapse or death from any cause. CSS was defined as the time from random assignment to the cancer-related death, unrelated death without evidence of disease was censored. Survival curves were estimated by the Kaplan-Meier method and differences among curves were evaluated by the log-rank test. A Cox proportional hazards model was used to adjust prognostic role of the factors considered for multiple baseline characteristics.
Comparison of proportions between groups was performed by use of a χ 2 test. BMI values were clustered according to World Health Organization ranges.
Results
Patients greater and less than or equal to 65 years were similarly distributed in the 2 arms of the original study: 92 women (34.8%) in the lymphadenectomy and 81 women (32.4%) in the no lymphadenectomy arm were >65 years. The younger patients were 172 (65.2%) in the lymphadenectomy and 169 (67.6%) in the no lymphadenectomy arm ( Table 1 ). FIGO stage, tumor grade, and BMI of trial patients divided by age are also reported in Table 1 . General characteristics of patients and their tumors were well balanced across the 2 age groups.
Demographic | Aged ≤65 (341 patients) | Aged >65 (173 patients) |
---|---|---|
Treatment arm | ||
Lymphadenectomy | 172 | 92 |
No lymphadenectomy | 169 | 81 |
FIGO stage | ||
IA | 4 | 4 |
IB | 140 | 54 |
IC | 117 | 67 |
IIA | 13 | 5 |
IIB | 11 | 14 |
IIIA | 18 | 10 |
IIIC | 29 | 14 |
IVC | 5 | 1 |
Missing | 4 | 4 |
Tumor grade | ||
1 | 32 | 6 |
2 | 193 | 105 |
3 | 113 | 56 |
Missing | 3 | 6 |
BMI | ||
<30 | 188 | 94 |
≥30 | 71 | 31 |
Table 2 shows the comparisons of 5-year OS and 5-year CSS by age groups and also by study arms within the 2 age strata. Patients >65 years, compared with the younger counterpart, showed statistically significant worse outcomes in both OS and CSS, and this poor prognosis was independent from whether they were submitted to lymphadenectomy or not ( Table 2 ).
Variable | n | 5-year OS | P value | 5-year CSS | P value |
---|---|---|---|---|---|
Aged ≤65 | 341 | 92.1% | < .001 | 93.8% | .003 |
Aged >65 | 173 | 78.4% | 83.5% | ||
Aged ≤65 | |||||
Lymphadenectomy | 172 | 91.3% | .63 | 92.9% | .60 |
No lymphadenectomy | 169 | 93.0% | 94.8% | ||
Aged >65 | |||||
Lymphadenectomy | 92 | 75.0% | .67 | 83.2% | .65 |
No lymphadenectomy | 81 | 82.5% | 83.6% |
Survival outcomes by nodal status in the 2 age groups are compared in Table 3 , where only patients enrolled in the lymphadenectomy arm (whose nodal anatomopathologic status was known) were considered.
Variable | n | 5-year OS | P value | 5-year DFS | P value | 5-year CSS | P value |
---|---|---|---|---|---|---|---|
Aged ≤65 | |||||||
Node negative | 148 | 94.4% | .009 | 91.2% | .001 | 96.3% | .002 |
Node positive | 24 | 74.3% | 70.4% | 74.3% | |||
Aged >65 | |||||||
Node negative | 81 | 75.7% | .55 | 78.8% | .53 | 83.6% | .58 |
Node positive | 11 | 74.1% | 75.0% | 83.3% |
Among women ≤65 years, node negative patients had 94.4% 5-year OS and 96.3% 5-year CSS vs 74.3% 5-year OS and 74.3% 5-year CSS for node positive patients ( P = .009 and P = .002, respectively), although among women >65 years, node negative patients had 75.7% 5-year OS and 83.6% 5-year CSS vs 74.1% 5-year OS and 83.3% 5-year CSS for node positive patients ( P = .55 and P = .58, respectively) ( Table 3 ).
Therefore the negative impact of nodal metastases was statistically significant only in younger (≤65 years) patients, although the worse survival of older patients was not significantly linked to nodal spread ( Table 3 ).
When considering BMI values as a continuous variable within the age strata, in the younger group (women ≤65 years), we found no significant association with OS or CSS, although in older women (women >65 years) higher values seem to be associated with CSS showing a hazard ratio of 1.10 (95% confidence interval [CI], 1.01–1.20; P = .03) and with OS with a hazard ratio of 1.11 (95% CI, 1.02–1.20; P = .01) ( Figures 1 and 2 ).
Throughout the trial, similar rates of adjuvant therapy were performed in younger and older patients: 31.4% vs 34.1% in women ≤65 years and >65 years, respectively ( P = .53), whereas adjuvant therapy was performed more frequently in patients with BMI ≤30 (37%) than in patients with BMI >30 (26.4%) ( P = .05).
Table 4 describes the site of disease recurrence by age (≤65 years vs >65 years). Distant relapse (lung, liver, bone, peritoneal) occurred in 10.5% vs 7.3% in >65 years and ≤65 years patients, respectively. Local (vaginal) recurrence occurred more frequently, even if not significantly, in older patients (4.6% vs 1.5%, respectively) as well. Retroperitoneal (nodal) relapse occurred in 1.8% and in 1.2% in ≤65 years and >65 years women, respectively.
Recurrence site | Aged ≤65 years (patients = 341) | Aged >65 years (patients = 173) | ||
---|---|---|---|---|
n | % | n | % | |
Recurrence | 39 | 11.4 | 28 | 16.2 |
Lung | 10 | 2.9 | 6 | 3.5 |
Bone | 5 | 1.5 | 2 | 1.2 |
Liver | 3 | 0.9 | 2 | 1.2 |
Intraperitoneal | 7 | 2 | 8 | 4.6 |
Vagina | 5 | 1.5 | 8 | 4.6 |
Lymph node | 6 | 1.8 | 2 | 1.2 |
Missing data | 4 | 1.2 | 2 | 1.2 |
Table 5 reports the univariate and multivariable analyses of CSS and OS by different prognostic factors in the whole population of the trial. The analyses showed that older age, higher tumor grade, and stage were significantly associated to a worse prognosis. Obese patients (BMI ≥30) showed a significantly worse OS only in multivariable analysis ( Table 5 ).
Variable | Univariate | Multivariable | ||||||
---|---|---|---|---|---|---|---|---|
Prognostic factor | CSS, HR (95% CI) | P value a | OS, HR (95% CI) | P value a | CSS, HR (95% CI) | P value a | OS, HR (95% CI) | P value a |
Treatment arm | ||||||||
No lymphadenectomy | 1.0 (referent) | .97 | 1.0 (referent) | .50 | ||||
Lymphadenectomy | 1.01 (0.55–1.85) | 1.20 (0.70–2.07) | ||||||
Age, y | ||||||||
≤65 | 1.0 (referent) | .004 | 1.0 (referent) | < .001 | 1.0 (referent) | .006 | 1.0 (referent) | .004 |
>65 | 2.46 (1.34–4.51) | 2.69 (1.57–4.63) | 2.57 (1.30–5.06) | 2.47 (1.33–4.58) | ||||
Tumor grade | ||||||||
1-2 | 1.0 (referent) | .01 | 1.0 (referent) | .01 | 1.0 (referent) | .01 | 1.0 (referent) | .03 |
3 | 2.15 (1.17–3.94) | 2.04 (1.19–3.50) | 2.11 (1.08–4.13) | 1.93 (1.05–3.58) | ||||
Tumor stage | ||||||||
I-II | 1.0 (referent) | .001 | 1.0 (referent) | .007 | 1.0 (referent) | .002 | 1.0 (referent) | .01 |
III-IV | 3.02 (1.59–5.76) | 2.44 (1.34–4.45) | 3.02 (1.49–6.11) | 2.48 (1.25–4.93) | ||||
BMI | ||||||||
<30 | 1.0 (referent) | .42 | 1.0 (referent) | .15 | 1.0 (referent) | .04 | ||
≥30 | 1.33 (0.66–2.68) | 1.58 (0.85–2.96) | 1.89 (1.01–3.57) | |||||
Adjuvant therapy | ||||||||
No | 1.0 (referent) | .21 | 1.0 (referent) | .54 | ||||
Yes | 1.48 (0.80–2.74) | 1.19 (0.68–2.07) |
Among women ≤65 years only tumor stage emerged as significant prognostic factor, even if patients who were submitted to adjuvant therapy showed worse CSS and OS only in univariate analysis ( Table 6 ).
Prognostic factor | Univariate | Multivariable | ||||||
---|---|---|---|---|---|---|---|---|
CSS, HR (95% CI) | P value a | OS, HR (95% CI) | P value a | CSS, HR (95% CI) | P value a | OS, HR (95% CI) | P value a | |
Treatment arm | ||||||||
No lymphadenectomy | 1.0 (referent) | .60 | 1.0 (referent) | .63 | ||||
Lymphadenectomy | 1.27 (0.52–3.07) | 1.21(0.54–2.72) | ||||||
Tumor grade | ||||||||
1-2 | 1.0 (referent) | .80 | 1.0 (referent) | .72 | ||||
3 | 1.12 (0.45–2.77) | 0.86 (0.36–2.02) | ||||||
Tumor stage | ||||||||
I-II | 1.0 (referent) | .002 | 1.0 (referent) | .009 | 1.0 (referent) | .002 | 1.0 (referent) | .009 |
III-IV | 4.11 (1.66–10.15) | 3.15 (1.34–7.42) | 4.11 (1.66–10.15) | 3.15 (1.34–7.42) | ||||
BMI | ||||||||
<30 | 1.0 (referent) | .54 | 1.0 (referent) | .87 | ||||
≥30 | 0.71 (0.23–2.18) | 0.93 (0.36–2.40) | ||||||
Adjuvant therapy | ||||||||
No | 1.0 (referent) | .02 | 1.0 (referent) | .04 | ||||
Yes | 2.87 (1.17–7.07) | 2.30 (1.02–5.16) |