Women with stage IA2 and IB1 cervical cancer typically are treated with radical hysterectomy and bilateral lymph node dissection, with or without accompanying oophorectomy or ovarian transposition. Radical hysterectomy includes removal of the upper third or half of the vagina as well as the parametrial tissue abutting the cervix. The removal of the parametrial tissue has been considered of paramount importance in the treatment of cervical cancer because tumor may spread to this area through direct microscopic extension or by tumor embolization from the primary lesion to lymph nodes embedded in the parametrial tissue.
Although the 5-year survival rate after radical hysterectomy is high, this surgical approach to early-stage cervical cancer is not without considerable morbidity. With radical hysterectomy, blood loss remains a major intraoperative complication: transfusion rates reach as high as 80%.1,2 In addition, a small but real risk of operative mortality exists with the surgery. Immediate postoperative complications include febrile morbidity, deep vein thrombosis, pulmonary embolism, and wound separations/dehiscence.3 Early-onset postoperative complications include fistula formation in as many as 4% of patients and prolonged bladder atony (lasting less than 6 months) in another 3–5% of women.4,5 Long-term complications can include bladder hypotonia requiring chronic self-catheterization, ureteral strictures, and chronic leg lymphedema.6,7
The majority of these intraoperative and postoperative complications are attributable to the parametrial resection as part of the radical hysterectomy. The parametrium is rich in vasculature and autonomic nerve fibers, and transection of these vessels and nerves can result in hemorrhage, devascularization of tissue leading to fistulae formation, or denervation of the bladder resulting in atony.8 Some surgeons advocate a “nerve-sparing” technique that may minimize some of these complications,9 but no prospective comparison study exists that shows that the “nerve-sparing” approach reduces morbidity without compromising mortality. In contrast to radical hysterectomy, simple hysterectomy spares the parametrial tissue and is associated with significantly less operative morbidity.
The objective of this study was to estimate the incidence of parametrial involvement in radical hysterectomy specimens from women with early-stage cervical cancer and to evaluate possible factors associated with parametrial spread. In addition, in an effort to estimate whether certain patients with early-stage cervical cancer are candidates for less radical surgery (eg, simple hysterectomy, simple trachelectomy, or cone biopsy), we analyzed the identified risk factors to develop an algorithm to predict which women would be at low risk for parametrial spread.
MATERIALS AND METHODS
This was a retrospective study approved by the institutional review board at The University of Texas M. D. Anderson Cancer Center, which waived the requirement for informed consent. We reviewed the medical records of all patients who underwent radical hysterectomy and pelvic lymphadenectomy for cervical cancer at M. D. Anderson Cancer Center from January 1, 1990, through December 31, 2006. We included patients with stages IA2–IB1 primary cervical cancer with squamous, adenocarcinoma, or adenosquamous lesions who underwent radical hysterectomy using laparotomy or laparoscopy. Variables examined included demographic factors, including age, race, body mass index (kg/m2), and smoking status, as well as pathologic factors, including stage, tumor size (2 cm or less or more than 2 cm), histology, grade, depth of invasion, lymphovascular space invasion, and status of pelvic lymph nodes. Patients for whom we were missing any data were excluded from analysis.
Pathologic specimens were processed as follows. At the time of gross examination, the parametrial margins were identified and submitted for microscopic evaluation. Next, if the tumor appeared to extend into the parametrium, contiguous full-thickness sections of the tumor in the cervix and parametrium were submitted. The parametrium then was searched for any palpable lymph nodes, which were placed in separate cassettes. Finally, the entire remaining parametrial tissue was submitted for microscopic examination. The assessment of lymphovascular involvement was made by microscopic evaluation of hematoxylin and eosin–stained slides.
The main outcome of interest was parametrial involvement in radical hysterectomy specimens. We first analyzed this endpoint as the dichotomous outcome of positive or negative parametrial involvement. We further explored this endpoint by examining the route of tumor spread into the parametrium, classified as 1) direct microscopic extension, 2) spread of tumor emboli to parametrial lymph nodes, 3) both direct microscopic extension and spread of tumor emboli to parametrial lymph nodes, and 4) tumor emboli in transit found within lymphovascular channels in the parametrial tissue. Finally, we examined the entire data set in an effort to identify a group of women with cervical cancer at low or no risk of parametrial involvement who would thereby be potential candidates for less radical surgery.
For the analysis of the dichotomous outcome of positive or negative parametrial involvement, normally distributed continuous variables were compared using Student’s t-test for independent samples. Nonnormally distributed continuous and dichotomous variable comparisons were performed using the Mann-Whitney U test and the χ2 test, respectively. Two-sided P-values were reported. Logistic regression analysis was performed to determine independent effects of pathologic variables on parametrial involvement. A P-value of less than .05 was considered to indicate statistical significance. SPSS 16 for Windows (SPSS Inc., Chicago, IL) was used for all statistical analyses. Owing to small numbers, statistical analysis was not performed on the data grouped by route of tumor spread.
Four hundred eight women with early-stage cervical cancer underwent radical hysterectomy at M. D. Anderson from 1990 to 2006. Thirty-four women were excluded from the analysis owing to one or more missing data points. Another 24 women who underwent radical hysterectomy for stage IA1 lesions also were excluded. The remaining 350 cervical cancer patients who underwent radical hysterectomy make up the study population. Demographic and pathologic factors for the entire cohort are shown in Table 1.
Overall, 27 (7.7 %) of the 350 women included had disease spread to the parametrium at the time of radical hysterectomy. Tumor size larger than 2 cm, higher grade, presence of lymphovascular space invasion, and positive pelvic nodes were associated statistically with parametrial involvement (Table 2). Women with grade 3 tumors had higher rates of parametrial involvement (12 %) compared with women with grade 2 tumors (6 %) or grade 1 tumors (0%) (P=.01). Viewed alternatively, 63% of women with parametrial involvement had grade 3 tumors and the remaining 37% of women with parametrial involvement had grade 2 tumors. Histologic subtype, however, had no significant effect on parametrial involvement.
Women with stage IB1 disease had an 8% risk of parametrial involvement, compared with 2% for women with stage IA2 disease (Table 2). Only 4% of women with tumors 2 cm or smaller had parametrial involvement, compared with 14% of women with tumors larger than 2 cm (P=.001). Lymphovascular space invasion also was associated significantly with parametrial spread; 12% of women with lymphovascular space invasion had parametrial involvement, compared with only 3% of those without lymphovascular space invasion (P=.002). Viewed alternatively, 81% of women with parametrial involvement also had lymphovascular space invasion in the primary tumor. The status of the pelvic lymph nodes also was correlated highly with parametrial involvement. Thirty-one percent of women with metastasis in the pelvic lymph nodes had parametrial involvement, compared with only 4% of those with negative pelvic lymph nodes (P<.001).
Of the 27 patients with disease found in the parametrium, 14 (52%) had only direct microscopic extension of disease, three (11%) had only disease spread to parametrial lymph nodes, six (22%) had both direct microscopic extension and disease spread to parametrial lymph nodes, and four (15%) had only tumor emboli within the lymphovascular channels in the parametrial tissue. Twelve (86%) of the women with direct microscopic extension of tumor into the parametrium had tumors larger than 2 cm in size. Lymphovascular space invasion was noted in 22 of the 27 patients with parametrial involvement: 11 (79%) of the 14 patients with direct microscopic extension, two (66%) of the three patients with tumor in parametrial nodes, five (83%) of the six patients with both direct microscopic extension and tumor in parametrial nodes, and all four (100%) of the patients with tumor emboli within the lymphovascular channels in the parametrium. Only five (36%) of the 14 women with direct extension had positive pelvic lymph nodes, whereas all three (100%) of the women with positive parametrial nodes and all six (100%) of the women with both direct extension and positive parametrial nodes had positive pelvic nodes. Only one (25%) of the four patients with tumor emboli in transit within lymphovascular channels in the parametrium had positive pelvic lymph nodes. Only one patient with stage IA2 cervical cancer had parametrial involvement, with disease found in a parametrial lymph node. This woman had a grade 2 adenocarcinoma with a depth of invasion of 4 mm and lymphovascular space invasion.
Multivariate analysis was performed for those risk factors that were most associated with parametrial involvement (Table 3). After controlling for confounding factors, positive pelvic nodes (odds ratio 6.47, P<.001) and tumor size more than 2 cm (odds ratio 2.44, P=.045) were the only factors that significantly predicted parametrial spread.
We were able to identify a group of women with cervical cancer at low risk for parametrial involvement (Tables 4 and 5). One hundred twenty-five women had tumors 2 cm or smaller without lymphovascular space invasion. None of the patients in this group had parametrial involvement. One patient (0.7%) in these 125 women had positive pelvic lymph nodes with negative parametrium. Therefore, for all women with squamous carcinomas, adenocarcinoma, or adenosquamous lesions, any grade, and a tumor 2 cm or smaller with no lymphovascular space invasion, the risk of parametrial involvement was 0% (0 in 125 patients).
Overall, we found parametrial involvement in 7.7% of women with early-stage cervical cancer undergoing radical hysterectomy. These findings are similar to those of previous studies, which have reported parametrial involvement in 3–11% of similar patients.10–14 In our study, tumor larger than 2 cm, higher histologic grade, presence of lymphovascular space invasion, and positive pelvic lymph nodes all predicted parametrial involvement. Furthermore, women with squamous, adenocarcinoma, or adenosquamous carcinomas, any grade, with small tumors (2 cm or less) and no lymphovascular space invasion had a 0% risk for parametrial involvement.
Most of the previous studies also found that larger tumor size, higher histologic grade, presence of lymphovascular space invasion, and positive pelvic nodes were associated significantly with parametrial involvement.10–14 In contrast to the other previously published studies and this current study, Covens and colleagues10 found a significant association between older age and parametrial involvement in their cohort of 842 women. However, the average age of women with parametrial involvement was 42 years and the average age of those without parametrial involvement was 40 years—only 2 years younger. The clinical significance of this statistically significant finding is likely negligible. Wright et al14 was the only group to find an association between histology and parametrial involvement. In their study, 3% of women with adenocarcinoma had parametrial involvement compared with 14% of women with squamous lesions and 26% with lesions categorized as “other” (P<.001). It is unclear, however, which tumor subtypes were categorized as “other,” and this group likely included high-risk histologies such as clear cell carcinoma, melanoma, and neuroendocrine tumors. Our study excluded clear cell carcinoma, melanoma, and neuroendocrine lesions but did include adenosquamous tumors, which we previously have reported are no more aggressive than squamous or adenocarcinoma lesions.15
Pelvic lymph node involvement previously has been reported to be a predictor of parametrial involvement.10,14 In our study, women with metastatic disease in the pelvic nodes were almost eight times as likely as women with negative pelvic nodes to have parametrial involvement (31% compared with 4%). Viewed alternatively, we found that 15 (56%) of the 27 women with parametrial involvement also had positive pelvic lymph nodes. However, 12 (44%) of the 27 women with parametrial involvement had negative pelvic lymph nodes. Strnad and colleagues16 found no discontinuous parametrial involvement in 133 women with low-risk cervical cancer and negative pelvic lymph nodes. Our findings confirm theirs: of the 12 women in our study with parametrial involvement and negative pelvic lymph nodes, nine had parametrial involvement through direct microscopic extension and the other three had in-transit tumor emboli. None of the 12 women with parametrial involvement and negative pelvic lymph nodes had positive parametrial nodes.
No studies have addressed the finding of positive pelvic lymph nodes with negative parametrial lymph nodes. Overall, 49 (14%) of the 350 patients in the current study had metastasis to the pelvic nodes. Of these 49 women, 34 (69%) had no metastatic disease in the parametrium. Anatomically, this is difficult to understand because many investigators, ourselves included, hypothesize that the lymph nodes in the parametrium are most likely the primary draining nodes for the cervix. Considered in light of the theory underlying sentinel node biopsy—namely, that tumor emboli deposit first in the primary draining nodal basin before metastasizing to second-echelon or nonsentinel nodes—our findings may be evidence for drainage of the primary cervical tumors directly to the pelvic lymph nodes. Another hypothesis to explain this phenomenon could be that some women might not have parametrial nodes or that very small parametrial nodes were not identified by the pathologist because only recently have the pathologists at our institution begun to submit the entire parametrium for pathologic sectioning.
Four women had parametrial involvement consisting of in-transit tumor emboli in the lymphovascular channels within the parametrium. It remains unclear what happens to these microscopic tumor emboli once they reach a draining lymph node—whether they implant and become a site of tumor metastasis or the body’s immune system clears this decidedly small-volume disease. The clinical significance of this finding, therefore, is largely unknown, and the clinically relevant percentage of women with parametrial involvement may be smaller than the 7.7% reported in this study.
Other authors have attempted to identify a subgroup of women with early-stage cervical cancer who are at low risk for parametrial involvement and therefore are potential candidates for less radical surgery (Table 6). The majority of these subgroups, however, have negative pelvic lymph node status as an inclusion criterion.10,13,14 Pelvic lymph node status would be unknown at the time of less radical surgery and pelvic lymphadenectomy, but women with metastatic disease in the pelvic nodes likely would undergo postoperative pelvic chemoradiation, thereby sterilizing the parametrial tissue left in situ.
Based on previous studies and our data, we have identified a low-risk subgroup of patients with cervical cancer who can be identified on the basis of data obtainable at the time of the loop electrosurgical excision procedure or cold knife conization. Such patients would be candidates for conservative surgery without removal of the parametrium. For that reason, our algorithm does not include pelvic node status. The criteria for our low-risk subgroup are squamous, adenocarcinoma, or adenosquamous lesions, any grade, with tumor size 2 cm or smaller and no lymphovascular space invasion. Of the 350 patients in our cohort, 125 women (36%) would have met these criteria, and none of these 125 women had parametrial involvement.
On the foundation of these findings, we have developed a prospective trial of conservative surgery for women with low-risk, early-stage cervical cancer. Eligible women will include those patients with squamous or adenocarcinoma lesions measuring 2 cm or less who have undergone either the loop electrosurgical excision procedure or cold knife conization with final pathology revealing no lymphovascular space invasion and negative margins. Women who meet these strict inclusion criteria and who desire future fertility will undergo bilateral pelvic lymphadenectomies followed by observation. Women who meet the inclusion criteria and have completed childbearing will undergo simple hysterectomy and bilateral pelvic lymphadenectomies. This international, multiinstitutional study currently is enrolling eligible patients.
1. Lentz SS, Shelton BJ, Toy NJ. Effects of perioperative blood transfusion on prognosis in early-stage cervical cancer. Ann Surg Oncol 1998;5:216–9.
2. Spirtos NM, Westby CM, Averette HE, Soper JT. Blood transfusion and the risk of recurrence in squamous cell carcinoma of the cervix: a gynecologic oncology group study. Am J Clin Oncol 2002;25:398–403.
3. Frumovitz M, dos Reis R, Sun CC, Milam MR, Bevers MW, Brown J, et al. Comparison of total laparoscopic and abdominal radical hysterectomy for patients with early-stage cervical cancer. Obstet Gynecol 2007;110:96–102.
4. Powell JL, Burrell MO, Franklin EW 3rd. Radical hysterectomy and pelvic lymphadenectomy. Gynecol Oncol 1981;12:23–32.
5. Samlal R, Velden J, Ketting B, González González D, Ten Kate F, Hart A, et al. Disease-free interval and recurrence pattern after the Okabayashi variant of Wertheim’s radical hysterectomy for stage IB and IIA cervical carcinoma. Int J Gyn Cancers 1996;6:120–7.
6. Magrina JF, Goodrich MA, Weaver AL, Podratz KC. Modified radical hysterectomy: morbidity and mortality. Gynecol Oncol 1995;59:277–82.
7. Michalas S, Rodolakis A, Voulgaris Z, Vlachos G, Giannakoulis N, Diakomanolis E. Management of early-stage cervical carcinoma by modified (Type II) radical hysterectomy. Gynecol Oncol 2002;85:415–22.
8. Ralph G, Winter R, Michelitsch L, Tamussino K. Radicality of parametrial resection and dysfunction of the lower urinary tract after radical hysterectomy. Eur J Gynaecol Oncol 1991;12:27–30.
9. Raspagliesi F, Ditto A, Fontanelli R, Zanaboni F, Solima E, Spatti G, et al. Type II versus Type III nerve-sparing radical hysterectomy: comparison of lower urinary tract dysfunctions. Gynecol Oncol 2006;102:256–62.
10. Covens A, Rosen B, Murphy J, Laframboise S, DePetrillo AD, Lickrish G, et al. How important is removal of the parametrium at surgery for carcinoma of the cervix? Gynecol Oncol 2002;84:145–9.
11. Kinney WK, Hodge DO, Egorshin EV, Ballard DJ, Podratz KC. Identification of a low-risk subset of patients with stage IB invasive squamous cancer of the cervix possibly suited to less radical surgical treatment. Gynecol Oncol 1995;57:3–6.
12. Steed H, Capstick V, Schepansky A, Honore L, Hiltz M, Faught W. Early cervical cancer and parametrial involvement: is it significant? Gynecol Oncol 2006;103:53–7.
13. Stegeman M, Louwen M, van der Velden J, ten Kate FJ, den Bakker MA, Burger CW, et al. The incidence of parametrial tumor involvement in select patients with early cervix cancer is too low to justify parametrectomy. Gynecol Oncol 2007;105:475–80.
14. Wright JD, Grigsby PW, Brooks R, Powell MA, Gibb RK, Gao F, et al. Utility of parametrectomy for early stage cervical cancer treated with radical hysterectomy. Cancer 2007;110:1281–6.
15. dos Reis R, Frumovitz M, Milam MR, Capp E, Sun CC, Coleman RL, et al. Adenosquamous carcinoma versus adenocarcinoma in early-stage cervical cancer patients undergoing radical hysterectomy: an outcomes analysis. Gynecol Oncol 2007;107:458–63.
16. Strnad P, Robova H, Skapa P, Pluta M, Hrehorcak M, Halaska M, et al. A prospective study of sentinel lymph node status and parametrial involvement in patients with small tumour volume cervical cancer. Gynecol Oncol 2008;109:280–4.
Figure. No caption available.
© 2009 by The American College of Obstetricians and Gynecologists.