Surgical management of cervical intraepithelial neoplasia in HIV-infected women

1. Introduction 

Surgical treatment eradicates cervical intraepithelial neoplasia (CIN) in around 90% of immunocompetent women whatever excisional technique is used [1]. Loop excision of the transformation zone (LLETZ) is considered the most appropriate technique [2]. Incomplete excision with positive margins has been reported as a risk factor of treatment failure [3], [4]. Lower efficacy below 50% of standard form of therapy for the treatment of CIN has been reported in HIV-infected women [5], [6]. As described in immunocompetent patients, positive margins are associated with a higher risk of treatment failure [7]. Two recent studies highlighted that HIV-infected women were twice more likely than HIV-negative matched women to have involved margins after conisation [8], [9].

The prognosis of patients with HIV/AIDS has improved markedly in the last decade with the widespread use of highly active antiretroviral therapy (HAART). Despite a protective effect against recurrent CIN following surgery, HAART does not clear human papillomavirus infection that play in important role in the pathogenesis of CIN [10]. Thus, the increase in life expectancy conferred by HAART may give time for CIN to recur or to progress to invasive cervical carcinoma. Improving the surgical management of CIN in HIV-infected women is therefore desirable.

In this study, we evaluated the efficacy of two excisional procedures, LLETZ and electrosurgical conisation, in obtaining complete excision of CIN in HIV-infected patients. The indication of surgical modalities depended on both the size and location of the lesion and on the length of the cervix. We compared the LLETZ and electrosurgical conisation with regard to the pathological specimens and the rate of complete excision.

2. Patients and methods 

All women attending the gynecology outpatient clinic of the HIV/AIDS departments of Hôpital Européen Georges Pompidou and Hôpital Cochin, Paris, between June 1993 and August 2006 were invited to participate in a prospective gynecological survey. The protocol was approved by the French National Agency for AIDS Research (ANRS). The eligibility criteria were documented HIV infection and informed consent. Among the 1129 eligible women, 1124 decided to enroll. Five declined for fear of loss of confidentiality. The study population comprised HIV-infected women undergoing surgical procedures at the Gynecology Department of Hôpital Cochin from June 1994 to August 2006.

At enrolment and every 6 months thereafter, the women had a pelvic examination, Papanicolaou smear and standardized colposcopic examination of the cervix, as previously described [11]. Patients with major colposcopic abnormalities or high-grade squamous intraepithelial lesions had a biopsy, unless they refused or did not return after receiving the result. Preoperative colposcopy was considered satisfactory when the entire transformation zone could be seen. It was considered unsatisfactory in case of poor visualisation of the transformation zone or when the upper limit of the transformation zone could not be visualised. All smears, biopsies and surgical specimens were examined by the same pathologist (Dr. Bergeron, Laboratoire Pasteur-Cerba, Cergy-Pontoise, France). The histologic categories were as follows: normal, metaplasia, low- or high-grade CIN, microinvasion, and carcinoma. Surgical specimen depth and width were measured in the pathology department after the specimen had been fixed in formalin. Margin status was considered positive if CIN was present histologically at the deep (endocervical) or external (exocervical) resection margin.

The follow-up protocol consisted of a cervical smear and colposcopy at a 6-month intervals. Further biopsies were prompted by abnormal cytology or the presence of an abnormal transformation zone. Endocervical curettage (ECC) was performed in case of unsatisfactory colposcopy.

3. Results 

During the study period, 80 HIV-infected women underwent 86 surgical treatments for cervical intraepithelial neoplasia or suspected cervical cancer. Repeat conisation was performed in 6 women because of recurrent high-grade CIN (5 cases) or suspected carcinoma (1 case). All the procedures were carried out by the same experienced operator (HF). Of the 80 women, 53% (42 women) were Caucasian and 71% (53 women) had been infected by HIV through heterosexual contact. At initial surgery their mean age was 35.9 years (31.6–40.3 years). Twenty-nine percent of the women were nulliparous (Table 1).

Table 1. Characteristics of HIV-infected women at first surgical treatment

		n=80	(%)
	Age
	Mean (years)	80	35.9
	IQR [25–75]		[31.6–40.3]
	Period
	1993–1996	14	(18)
	1997–2000	21	(26)
	2001–2006	45	(56)
	Ethnic origin
	Caucasian	42	(53)
	Sub-Saharan Africa	25	(31)
	North Africa	13	(16)
	Transmission group
	Heterosexual contact	53	(71)
	Injection drug users	21	(28)
	Blood contaminated	1	(1)
	Missing data	5
	Parity
	0	17	(29)
	≥1	42	(71)
	Missing data	21

The characteristics of the referral biopsies and the preoperative colposcopic findings are shown in Table 2. Preoperative biopsy showed high-grade CIN in 78 cases, low-grade CIN in 3 cases, and suspected cancer in 5 cases. Two of the three women with low-grade CIN had been diagnosed with high-grade CIN by prior biopsy. In three women, both the preoperative Pap smear and the biopsy findings were suggestive of carcinoma, whereas in two other women the preoperative Pap smear showed features of carcinoma and biopsy showed high-grade lesions. Preoperative colposcopy failed to visualize the entire transformation zone in 39% of cases, and showed that 93% of the lesions had endocervical extension.

Table 2. Characteristics of referral biopsy, preoperative colposcopy by treatment group

		Totala		Treatment groupb
				LLETZ		Surgical conisation		P
		n=86	(%)	n=30	(%)	n=56	(%)
	Referral biopsy
	Low-grade CIN	3	(3)	3	(100)	0	(0)	0.05
	High-grade CIN	78	(91)	26	(33)	52	(67)
	Cancer Suspicion	5	(6)	1	(20)	4	(80)
	Preoperative colposcopy
	Satisfactory	49	(61)	21	(43)	28	(57)	0.01
	Unsatisfactory	31	(39)	5	(16)	26	(84)
	Missing data	6
	Location of lesion on preoperative colposcopy
	Strictly endocervical	14	(20)	6	(43)	8	(57)	0.23
	Both endo+exocervical	51	(73)	14	(27)	37	(73)
	Strictly exocervical	5	(7)	3	(60)	2	(40)
	Missing data	16

a Vertical percent. b Horizontal percent.

LLETZ was performed in 30 cases and electrosurgical conisation in 56 cases. As expected, unsatisfactory preoperative colposcopy was significantly more frequent in the electrosurgical conisation group than in the LLETZ group (84% versus 16%, p=0.01). Sutures were required for hemostasis significantly more frequently during electrosurgical conisation than during loop excision (42% and 11%, respectively, p=0.005). There were no intraoperative complications in either subgroup and no women were readmitted for secondary haemorrhage. All loop excision and electrosurgical conisation specimens were suitable for review. The margins were assessable in all but one case.

Table 3 shows the characteristics of the surgical specimens and the extent of lesions at the endocervical and ectocervical margins according to the treatment group. Microinvasive cancer was diagnosed in two cases (one patient had a preoperative Pap smear with features of carcinoma and a biopsy showing high-grade lesions) and adenocarcinoma in two cases. High-grade CIN, including a case of adenocarcinoma in situ, was diagnosed in 75 cases. Low-grade CIN or metaplasia was diagnosed in 7 cases. Endocervical CIN was observed in 91% of surgical specimens. The median length and width of the cone specimens was significantly higher after electrosurgical conisation (p<0.001). Half the specimens obtained by electrosurgical conisation were more than 13mm deep.

Table 3. Characteristics of surgical specimens by treatment group

		Total		LLETZ		Surgical conisation
		n=86	(%)	n=30	(%)	n=56	(%)
	Histological specimen
	Metaplasia	2	(2)	0	(0)	2	(4)
	Low-grade CIN	5	(6)	4	(13)	1	(2)
	High-grade CIN	75	(87)	26	(87)	49	(87)
	Microinvasive cancer	2	(2)	0	(0)	2	(4)
	Adenocarcinoma	2	(2)	0	(0)	2	(4)
	Location of CIN on histological specimen
	Strictly endocervical	26	(34)	9	(35)	17	(33)
	Both endo+ectocervical	44	(57)	13	(50)	31	(61)
	Strictly exocervical	7	(9)	4	(15)	3	(6)
	Missing data	9
	Histological specimen median length (mm)
	Median	78	12.0	25	10.0	53	13.0
	IQR [25–75]		[10.0–15.0]		[8.0–11.0]		[10.0–18.0]
	Missing data	8
	Histological specimen median width (mm)
	Median	78	23.5	25	20.0	53	25.0
	IQR [25–75]		[20.0–27.0]		[17.0–21.0]		[22.0–30.0]
	Missing data	8
	Excision status
	Complete excision	63	(73)	23	(77)	40	(71)
	Endocervical margin involved	19	(22)	6	(20)	13	(23)
	Both margin involved	1	(1)	0	(0)	1	(2)
	Exocervical margin involved	2	(2)	1	(3)	1	(2)
	Missing data (margin involved)	1	(1)			1	(2)

The cervical lesions were completely excised in 63 (73%, 95% confidence interval, CI: 63–82%) of the 85 cases with known margin status. There was no difference between the treatment groups in the rate of lesion clearance (77% (95% CI: 62–92) and 71% (95% CI: 60–83), after LLETZ and electrosurgical resection respectively). Residual disease was mostly located in the endocervical portion of the specimens (20/22, 91%).

A logistic regression analysis was conducted with the aim to identify variables that were significant predictors of positive margin. The following variables were assessed: a variable combining surgical modalities and preoperative colposcopy, and the specimen length. In univariate analysis, none of the test variables appeared predictive of positive margins (Table 4).

Table 4. Analysis of factors associated with positive margin (n=80a)

			Positive margin		Univariate logistic model
		N	n	(%)	OR	[95% CI]	P
	Preoperative colposcopy and treatment group
	Colposcopy satisfactory and surgical conisation	28	6	(21)	1		0.52
	Colposcopy satisfactory and LLETZ	21	4	(19)	0.9	[0.2 3.6]
	Colposcopy unsatisfactory and surgical conisation	26	9	(35)	1.9	[0.6 6.5]
	Colposcopy unsatisfactory and LLETZ	5	2	(40)	2.4	[0.3 18.1]
	Histological specimen length
	<15mm	46	10	(22)	1		0.26
	≥15mm	30	10	(33)	1.8	[0.6 5.1]
	Missing data	4
	Location of CIN on Histological specimen
	Strictly endocervical	26	10	(38)	1		0.06
	Strictly exocervical or both endo+exocervical	49	9	(18)	0.4	[0.1 1.1]
	Missing data	5

a Missing data at preoperative colposcopy=6.

Fourteen of the 80 women (17.5%) were lost to follow-up. None of the three women with low-grade CIN who were followed up had persistent/recurrent CIN. Among the 61 women who had high-grade CIN on histological examination and who attended at least one follow-up visit, 30 had persistent/recurrent lesions, with low-grade CIN in 16 cases and high-grade CIN in 14 cases. One patient with persistent abnormal features at Pap smear and unsatisfactory colposcopy had a hysterectomy 10 months after electrosurgical conisation. Malignant cells were not found in the hysterectomy specimen; however a vaginal carcinoma was diagnosed nine months after hysterectomy. One of the two patients with microinvasive cancer remained disease-free seven years after electrosurgical conisation with negative margins, while the second patient was lost to follow-up after surgical conisation with positive margins. The patients with adenocarcinoma received appropriate treatment.

We did not observed progression to invasive cervical cancer during follow-up. Late complications such as cervical stenosis or unsatisfactory colposcopy were not observed.

4. Comment 

Resection of CIN is incomplete in more than 40% of HIV-infected women [8], [9] and failure rate as high as 56% have been observed after LLETZ [5]. This high rate of positive margin might be related to endocervical extension of lesions. Wright et al. [5] reported a 15% rate of positive endocervical curettage among HIV-infected women despite satisfactory preoperative colposcopy.

Our study endorses this finding, since we observed CIN involving the endocervix in 91% of cases on histological examination of the surgical specimens. Furthermore, deep endocervical extension of the lesion was frequent in our study, as shown by 10 of 30 histological specimens more than 15mm long with incomplete endocervical resection.

Current treatment for CIN by excisional methods consists of various techniques including knife cone biopsy, electrosurgical conisation and LLETZ. Knife cone biopsy is usually preferred for endocervical extension of the lesion in spite of potential peri-operative homeostasis difficulties and risk of cervical stenosis and unsatisfactory colposcopy during follow-up [1]. For these reasons, we considered electrosurgical conisation for CIN with endocervical extension, which alike cold knife conisation, yields broad and deep cones [12]. We used an electrode powered by a high-frequency electrosurgical generator, which allows excision of deep cones of cervix. The median length of histological specimen was 13.0mm (10.0–18.0) in our study. Excision was complete in 71% of the cases. Both short- and long-term morbidity were avoided. Primary haemorrhage was rare, and suturing was necessary in less than 50% of cases. This procedure was not associated with cervical stenosis or unsatisfactory colposcopy during follow-up. It also allowed repeated excisional procedures to monitor recurrence, thus preventing hysterectomy in these patients who desire fertility.

LLETZ is by far the most popular procedure. Poor result of loop electrosurgical excision was previously reported in HIV-infected women, and the relevance of this modality for treating CIN in HIV-positive women was questioned [5]. LLETZ was nevertheless the surgical procedure more frequently used in HIV-infected women. We restricted the use of this procedure to patients in whom either the lesion was small and fully visible on preoperative colposcopy, or in case of unsatisfactory colposcopy, whose uterine cervix was long and narrow. LLETZ had been the chosen procedure in only one-third of our patients. Restricting LLETZ indication to these situations, we actually obtained complete excision in 77% of CIN indicating high therapeutic efficiency of this modality when indication is defined according to colposcopic findings. This high efficacy contrasts with previous studies that reported positive margin in more than 50% cone biopsies of HIV-infected women [6], [8], [9]. We avoided combination of LLETZ with additional apical excision (‘top hat’ or ‘second pass’) since previous report indicated that the second pass was associated with higher rate of cervical stenosis [13].

In our study, compared with electrosurgical conisation, the LLETZ did not have a higher rate of involved margin (p=0.52), which is actually related to the fact that LLETZ was only performed under specific circumstances. Our failure to detect a correlation between the length of the cone biopsy and the presence of disease at the margins might be related to small sample size and further work is warranted to confirm this finding.

In this setting of high risk of recurrence, post-operative management is crucial. We did not observe cervical stenosis. Women were followed-up with a systematic colposcopy and Pap smear or ECC in case of unsatisfactory colposcopy at a 6-month intervals. No adenocarcinoma was identified during follow-up.

Incomplete excision of CIN exposes women to a substantial risk of post treatment disease. This study is the first, to our knowledge, to report on a high rate of 73% of complete excision in the setting of HIV. Even if endocervical extension of CIN is frequent in HIV-infected women, a similar rate of positive margin was observed whether electrosurgical conisation or LLETZ was used. This could only be achieved by adapting the treatment modalities to the location of the lesion evidenced on colposcopy and to the shape of the cervix.