ORIGINAL RESEARCH ARTICLE: CERVIX AND HPV

Recurrence in Cervical High-Grade Squamous Intraepithelial

Lesion: The Role of the Excised Endocervical Canal Length—
Analysis of 2,427 Patients
Andresa Ines Foggiatto, MD,1 Newton Sergio de Carvalho, PhD, MD,1
Fernanda Villar Fonseca, PhD, MD,2,3 and Carlos Afonso Maestri, PhD, MD2,3

neoplasia (CIN). Richart (1967) classified CIN as 1, 2, and 3 de-

Objective: This study aimed to evaluate the influence of the excised canal pending on the proportion of epithelial thickness that presents ma-
length on relapse rates of cervical high-grade squamous intraepithelial le- ture and differentiated cells. The Bethesda System (1992) created
sion (HSIL) treated by loop electrosurgical excision procedure and to find the term squamous intraepithelial lesion (SIL), which comprised
a cut-off point, above which lower recurrence rates could be observed, with low-grade and high-grade (HSIL) lesions. Low-grade lesions in-
low probability of compromising future obstetric outcome, and the rela- clude flat condylomatous changes (HPV) and low-grade CIN
tionship with other individual factors related to HSIL recurrence. (CIN1), whereas high-grade intraepithelial lesions (HSIL) com-
Method: This was a retrospective cohort study of 2,427 women diag- prise CIN 2 and 3.3,4
nosed with cervical intraepithelial neoplasia CIN2+ who underwent cervi- Most low-grade CIN regress in relatively short periods or do
cal conization using the high-frequency loop electrosurgical excision pro- not progress to high-grade lesions. High-grade CINs are more
cedure surgery technique, to analyze the role of endocervical canal length likely to progress to invasive cancer.3
associated with individual factors in the recurrent disease after CIN2+ treat- Studies show that persistence of high-risk oncogenic HPV
ment and determine a cut-off point for the excised canal length needed to may determine progression to more severe stages of cervical dis-
decrease the risk of disease relapse. ease. The risk for disease progression seems to be associated with
Results: In 2,427 cases, the relapse rate of HSIL treated was 12%. Com- the host HPV cycle.3,5–9 The natural history of HPV in the human
promised margins of conization, HIV+, and endocervical canal length were host has some points not yet fully elucidated. Human papillomavi-
related directly to relapses ( p < .001). The cut-off point, by receiver oper- rus infection is very common among young women during sex
ating characteristic curve, to calculate the endocervical canal length related life; approximately 70% of women will have contact and/or tran-
to relapses was 1.25 cm of canal excised. Canal length of less than 1.25 cm sient infection because of it. Despite these high rates of infection,
increased the recurrence rate 2.5 times. Compromised margins and HIV+ few infected women will develop cervical cancer.8
increased recurrence rates by more than 5 times. The treatment of high-grade lesions, when chosen, can be
Conclusion: Cervical HSIL recurrence was directly related to the endo- performed by destructive methods such as cauterization or exci-
cervical canal length: excised canal length of 1.25 cm or more decreases re- sion, with preference for excision by obtaining a surgical speci-
currence rate; HIV and compromised margins increase the chance of recur- men that rules out the possibility of invasion. The most commonly
rence by more than 5 times. used surgical technique is conization, which can be performed by
Key Words: high-grade squamous intraepithelial lesion, conization, cold scalpel method or by electrosurgery technique: the large loop
loop electrosurgical excision procedure, recurrence, excision of the transformation zone and loop electrosurgical exci-
cervical intraepithelial neoplasia, canal length sion procedure (LEEP), also known in Brazil as excision
transformation zone.10
(J Low Genit Tract Dis 2023;27: 1–6)
Studies consider residual disease that were identified during
the first year after conization or recurrence when it happens after
C ervical cancer is a major public health problem in Brazil and
in developing countries and is associated with persistent in-
fection with human papillomavirus (HPV) oncogenic types, espe-
this period.11,12
Some known factors related to residual neoplasia or recur-
rence after conization are the degree of CIN, compromised exci-
cially 16 and 18, responsible for approximately 70% of
sion margins and glandular involvement in the conization product,
cervical cancers.1,2
age, race, number of pregnancies and deliveries, socioeconomic
Cervical malignancies are preceded by a long phase of
status, smoking, marital status, nonvaccination against HPV, and
preinvasive disease, jointly denominated as cervical intraepithelial
presence of high-risk HPV.5–15
1
The CIN relapse rates after conization by LEEP range from
Gynecology and Obstetrics Department, Federal University of Parana, Curi-
tiba, Parana, Brazil; 2Gynecology Department, Positivo University, Curitiba,
5% to 30%, and patients need follow-up and sometimes retreatment
Parana, Brazil; and 3Cervical Pathology Department, Erasto Gaertner Cancer when lesions are identified.14–20 Major relapse-related complica-
Center Hospital, Curitiba, Parana, Brazil tions are compromised margins and associated HIV infection.10
Reprint requests to: Fernanda Villar Fonseca, PhD, MD, R. Dr Ovande do Some studies have shown that the main complications related to
Amaral, 201-Jardim das Américas, Curitiba, Paraná, Brazil. CEP
81520-060. E-mail: [email protected]
the removal of the canal length would be an increase in future ob-
The authors have declared they have no conflicts of interest. stetric comorbidities such as premature delivery, premature
Andresa Ines Foggiatto ORCID: 0000-0002-4733-0256. amniorrhexis, and low birth weight.21,22
Newton Sergio de Carvalho ORCID: 0000-0001-7561-4566. On review of the literature, no studies were found that evalu-
Fernanda Villar Fonseca ORCID: 0000-0002-0162-9893.
Carlos Afonso Maestri ORCID: 0000-0003-0200-9802.
ated a cut-off point of the length of the resected canal with recur-
Ethics committee status: approved on March 28, 2017. rence rates. On the other hand, there are many studies that demon-
This study evaluated the predictive factors that could influence persistence or strate the importance of canal length in those who underwent sur-
recurrence of high-grade squamous intraepithelial lesion after treatment gery with later pregnancy. In these studies, there are data that
with loop electrosurgical excision procedure in 2,427 women, including the
role of endocervical canal length.
report that greater length of the removed canal related to the occur-
© 2022, ASCCP rence of premature birth.23,24 Thus, the objective here is to evalu-
DOI: 10.1097/LGT.0000000000000708 ate the influence of the excised canal length on relapse rates of

Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023 1

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.

Foggiatto et al. Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023

disease and to find a cut-off point above which lower recurrence - Exclusion criteria: colposcopy-directed biopsy showing a dif-
rates could be observed associated with low probability of ferent report of CIN 2 or 3 (excluding reports of cervicitis and
compromising the obstetric outcome. microinvasive cancer); patients with clinical follow-up shorter
than 18 months.

MATERIALS AND METHODS

This is an observational, descriptive, retrospective, longitudi- Data Collection. Epidemiological data were collected only by
nal study of women diagnosed with CIN2+ (CIN 2 or 3/HSIL) the researcher group through a review of the electronic medical
who underwent cervical conization using the high-frequency records of the patients included in the study. Epidemiological
LEEP surgery technique at the Cervical Pathology Service at data (age, smoking, HIV seropositivity, clinical follow-up after
Erasto Gaertner Cancer Center in Curitiba, Paraná, Brazil, from conization), histological data (histological diagnosis of cervical
January 2009 to December 2016. The project was approved by biopsy), and data related to the treatment performed (conization
the research ethics committee. technique, conization analysis of the product margins, piece
volume, and canal length) were collected.
Patient Selection, Data Collection, and Sample
Study Design. The global sample of patients selected by the
Patient Selection. From January 2009 to December 2016, inclusion and exclusion criteria resulted in 2,427 patients. This
4,016 patients who underwent the cervical conical resection
can be seen in Flowchart 1 (global patient sample).
process were included. In this population, the following criteria
were applied:
Conization Technique and Canal
- Inclusion criteria: patients treated at the Cervical Pathology Length Assessment
Service, with cytohistological diagnosis through Pap smear The conization technique was performed using LEEP in all
and subsequent colposcopy-directed biopsy confirming CIN 2 cases, according to the protocol of the Cervical Pathology Service,
or 3. This cohort was submitted to a conical cervical in which lesions considered too large and/or deep were previously
resection procedure. determined for the classical technique. The procedure was

FIGURE 1. Flowchart 1 - Disease cut-off point and recurrence analysis.

2 © 2022, ASCCP

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.

Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023 Risk Factors to HSIL Recurrence

TABLE 1. Global Sample Epidemiological Data (n = 2,427 TABLE 3. Relationship of Data Related to the Surgical Specimen
Cases) With HSIL Relapses/Recurrence

Age, y 40 ± 11 (18–93) Recurrence

Parity 2 ± 1.8 children (0–13) NO YES
Smokinga Yes 668 (27.5%) Average DP Average DP p
No 1,755 (72.4%)
HIVa Yes 56 (2.6%) Excised canal length, cm 2.75 0.78 2.57 0.86 0.01
No 2,317 (97.3%) Total volume of the 5.28 4.44 4.88 3.63 0.16
excised piece, cm3
Biopsy histology CIN 2 1,288/1,699 (75.8%)
CIN 3 411/1,699 (24.2%) Data source: the authors (n = 2,427 cases).

Data source: the authors.

Global sample: n = 2,427 cases.
a
In 0.1% of cases, these data were not accessible, as this was a retrospective
relapse, a descriptive analysis was performed. Data were
study. presented by distribution of absolute and relative fre-
quencies, and comparisons between relapses and
nonrelapses were performed with the χ 2 test.
performed using the LEEP classical technique described by 3. Cut-off point for analysis: the receiver operating characteris-
Prendiville et al.24 in 1989, under local anesthesia in all cases. tic (ROC) curve was used to estimate a cut-off point where the
Surgery was performed in an outpatient setting, and the pa- chance of relapse could be higher.
tients returned for postoperative review within 20 days and there- 4. Association: possible associations between canal length and
after, semestral clinical follow-up. relapse were verified by binary logistic regression. For the ad-
The canal length was measured with a millimeter ruler by cal- justed analyses, factors such as age, parity, and smoking
culating the upper edge of the outer margin with the lower edge. were included.
The volume of uterine cervix removed was calculated by mul-
tiplying the length by the height by the width of the removed piece.

Statistical Data Analysis RESULTS

Statistical procedures were performed using software SPSS From January 2009 to December 2016, 4,016 conizations
version 24.0 (IBM, Armonk, NY) and STATA version 13.0 were performed at the Cancer Center. After exclusion criteria were
(StataCorp LLC, College Station, TX), with p set at 0.05, follow- applied, 2,427 patients were selected.
ing these strategies: Of the 2,427 patients, 311 (12.8%) relapsed and 2,116
(87.2%) did not relapse, as shown in Flowchart 1 (Figure 1).
1. Data normality: verified by the Kolmogorov-Smirnov test. The average age was 40 ± 11.5 years (18–93), the parity was
2. To observe if there was any relationship between the 2 ± 1.8 children (0–13). Smoking was present in 668 women and
length of the cervical canal removed and the chance of HIV was positive in 2.6% (56) of the patients.
Free margins were found in 87.3% (2,119) of the cases and
compromised in 12.6% (308). Of the compromised margins,
TABLE 2. Relationship of Data Related to the Surgical Specimen 42.85% were ectocompromised, 43.83% endocompromised, and
With HSIL Relapses/Recurrence ( p Value to Recurrence) in 12.33% of them, both margins were compromised. In patients
Variable Results p
Age, y 40 ± 11 (18–93) 0.01 TABLE 4. Comparative Data Between Cases of Recurrence and
Parity 2 ± 1.8 children 0.06 Nonrecurrence in Relation to the Canal Length Excised
(0–13)
Smokinga Yes 668 (27.5%) 0.45 Recurrence Recurrence rate
Excised canal
No 1,755 (72.4%) length Yes No (%)
HIVa Yes 56 (2.6%) 0.001 <1.25 cm 17/61 44/61 28%
No 2,317 (97.3%) 1.25–2 cm 64/390 326/390 16%
Biopsy histology CIN 2 1,288/1,699 (75.8%) 0.34 2–2.5 cm 71/663 592/663 11%
CIN 3 411/1,699 (24.2%) >2.5 cm 160/1,313 1,154/1,313 12%
Conization Free 2,116 (87.3%) 0.001
margins
Note: Using ROC curve, 1.25 cm was the cut-off point where the
Involved 308 (12.6%) chance of recurrence rates can be higher.
Average length 2.7 cm ± 0.8 cm 0.01 Recurrence rate
of the
excised canal No Yes p All
Average volume 5.2 cm3 ± 4.3 cm3 0.16 n % n % N %
of the Excised canal <1.25 cm 44 1.8 17 0.7 0.001 60 2.5
excised piece length
a
≥1.25 cm 2,072 85.4 295 12.2 2,367 97.5
In 0.1% of cases, these data were not accessible, as this was a retrospec-
tive study. Data source: the authors.

© 2022, ASCCP 3

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.

Foggiatto et al. Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023

FIGURE 2. The ROC curve for all sample/cut-off point for the canal length from which the HSIL recurrence rate increases.

with compromised margins, relapses occurred in 116 (37.7%) and related to the chance of relapse. All these data can be seen in
in 196 (9.2%) in those with free margins, and it was statistically Table 2.
significant ( p < .001). Recurrence occurred in 311 of the 2,427 patients, repre-
The epidemiological data and those related to the product of senting 12.8%. The average time to relapse was 13 months
conization can be seen in Table 1. ±12.5 months. Relapse was 52% at 10 months, 62.5% at
Epidemiological data show that age and parity were sta- 12 months, and 80.4% at 18 months, considering that only pa-
tistically significant to disease recurrence, whereas smoking tients who had follow-up beyond 18 months were included.
was not. The presence of HIV was demonstrated to be highly The positivity of HIV was a significant factor for recurrence
significant in association with relapsing. Regarding the ex- ( p < .001), with relapse occurring in 12% of seronegative and
cised specimen data, it was observed that the biopsy histolog- 43% of seropositive patients.
ical report and volume were not related to recurrence rates. On The study showed that there was a significant difference
the other hand, margin impairment and canal length were directly between relapses and nonrelapses in relation to canal length,

TABLE 5. The Recurrence Probability (OR) for Treated HSIL in Relation to Canal Length, HIV Seropositivity, and Margins of the
Conization Product

Raw analysis Adjusted analysis

OR 95% CI p OR 95% CI p
Canal length ≥1.25 cm 1 — — 1 — —
<1.25 cm 2.55 1.42 – 4.58 0.002 2.40 1.26 – 4.56 0.007
HIV Negative 1 — — 1 — —
Positive 6.67 3.93 – 11.29 0.001 5.65 3.20 – 9.97 0.001
Compromised margin Free 1 — — 1 — —
Compromised 0.99 4.55 – 7.87 0.001 5.83 4.38 – 7.76 0.001
Adjusted analysis for all factors: p < .05.
OR indicates odds ratio.
Data source: the authors.

4 © 2022, ASCCP

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.

Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023 Risk Factors to HSIL Recurrence

although the analysis of the total volume of the piece did not defined cut-off point, we may have contributed to the reduction of
show such association, as can be seen in Table 3. obstetric comorbidities and lower recurrence rates.
As evidenced in the literature that the chance of relapsing is In addition to these main data found on the minimum length
higher in patients with compromised margins and HIV,11 based of the excised canal, we found a direct relationship between HIV
on the analysis of the canal length as a function of the relapse rate, and margin impairment (6.67 and 5.99 times, respectively) and re-
these patients were excluded, comprising a population that was currence rate, demonstrating the credibility of the data presented
called “purified population.” It could be argued that the possibility in agreement with the literature regarding these factors (HIV and
of greater margin compromise would be associated with the margins impairment) and relapse rates.
shorter length of the removed canal. After statistical analysis, it Within the same reasoning, it could be inferred that in
was shown that there was a correlation between excised cervical manufacturing the handles used in these procedures, it ideally
canal length and chance of relapsing. could be calibrated for confection higher than 1.25 cm in relation
We observed that the resected volume of the piece was not to their height.
statistically significant; therefore, we focused on the length of At this time, after data analysis, the study concluded that:
the canal, and that was the main object of the study analysis as
shown in Table 4. 1. The excised canal length is directly related to the relapse rates;
To evaluate if there was a certain cut-off point of the excised 2. An excised canal length of 1.25 cm or longer was associated
canal length that was associated with higher chance of recurrence, with a lower recurrence rate;
the data were analyzed using an ROC curve, as can be seen in 3. Human immunodeficiency virus and compromised mar-
Figure 2. The graph shows the ROC curve of the general popula- gins presented with 6.67 and 5.99 times higher chance of
tion showing that 1.25 cm or longer of canal resection would have recurrence, respectively.
a lower chance of recurrence.
In an attempt to evaluate the chance of relapse in relation to
the factors listed in Table 5, we observed the odds ratio of each
factor as a function of recurrence. The study demonstrated that
the presence of compromised margins increases the risk of relapse REFERENCES
by 5.99 times ( p < .001; 95% CI = 4.55–7.87). The presence of
1. Ministério da Saúde; Instituto Nacional do Cancer (Brasil). Estimativa
HIV increases the risk of relapse by 6.67 times ( p < .001; 95%
2018. Incidência do Câncer no Brasil. Rio de Janeiro, Brazil: INCA; 2016.
CI = 3.93–11.29). When the length of the excised canal is shorter
than 1.25 cm, the risk of recurrence increases by 2.55 times (95% 2. Kolf C. QnAs with Harald zur Hausen. Proc Natl Acad Sci U S A 2012;
CI = 1.42–4.58). 109:1378.
As limiting factors, we suggest that this was a retrospective 3. International Agency for Research on Cancer. Monographs of Evaluation
study, in which the pieces were evaluated and the canal length of Carcinogenetic Risks to Humans. Lyon, France: IARC; 2007;90:1–689.
was measured without a planned association with recurrence rates. 4. The revised Bethesda System for reporting cervical/vaginal cytologic
Moreover, the fact that we did not link to the location of the le- diagnoses: report of the 1991 Bethesda workshop. Acta Cytol 1992;36:
sions, if ectocervical or endocervical (intracanal), may suggest 273–6.
to us that in the latter, canal length value may be underestimated. 5. Richart RM. Natural history of the cervical intraepithelial neoplasia. Clin
The procedures were performed by various health profes- Obstet Gynecol 1967;10:748–84.
sionals, and the anatomopathological specimen was evaluated by
6. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is
different sets of pathologists; this interoperator/interobserver
a necessary cause of invasive cervical cancer worldwide. J Pathol 1999;
“bias” is likely overcome by sample size. 189:12–9.
7. Villa LL. Human papillomaviruses and cervical cancer. Adv Cancer Res
1997;71:321–41.
DISCUSSION AND CONCLUSIONS 8. Moscicki AB, Shiboski S, Broering J, et al. The natural history of human
Because of the dilemma between the lower chance of recur- papillomavirus infection as measured by repeated DNA testing in
rence of cervical lesion (CIN) and the possibility of increasing adolescent and young women. J Pediatr 1998;132:277–84.
the rate of prematurity, the excision of the canal should not be less 9. Muñoz N, Bosch FX. HPV and cervical neoplasia: review of case-control
than that sufficient to prevent recurrence of the lesion (high-grade and cohort studies. IARC Sci Publ 1992;251–61.
CIN) and not greater than that which could increase the risk of 10. Fonseca FV, Tomasich FD, Jung JE, et al. The role of P16 ink4a and P53
prematurity rate. immunostaining in predicting recurrence of HG-CIN after conization
Therefore, based on this rationale, the goal of this study and treatment. Rev Col Bras Cir 2016;43:35–41.
as-found novelties has shown the direct relationship between the 11. Gonzales DIJr., Zahn CM, Retzloff MG, et al. Recurrence of dysplasia after
measurement of the excised (resected) canal length and the recur- loop electrosurgical excision procedures with lon-term follow-up. Am J
rence after conization. Obstet Gynecol 2001;184:315–21.
Among the excised canal measurements, the cut-off point of
12. Milojkovic M. Residual and recurrent lesions after conization for cervical
1.25 cm was obtained, below which recurrence rates increased by
intraepithelial neoplasia grade 3. Int J Gynaecol Obstet 2002;76:49–53.
2.55 times, and these data have not yet been found in the literature.
Among the perspectives, it is important to highlight that be- 13. Maluf PJ, Adad SJ, Murta EF. Outcome after conization for cervical
sides countless variables already shown in the literature, which in- intraepithelial neoplasia grade III: relation with surgical margins, extension
crease recurrence after conization, the length of the canal also pre- to the crypts and mitoses. Tumori 2004;90:473–7.
sents this assertion. 14. Alonso I, Torné A, Puig-Tintoré LM, et al. High-risk cervical epithelial
This study opens perspectives so others can evaluate this neoplasia grade 1 treated by loop electrosurgical excision: follow-up and
finding in other samples and other populations (the canal length value of HPV testing. Am J Obstet Gynecol 2007;197:359.e1–6.
cut-off point). 15. Bae JH, Kim CJ, Park TC, et al. Persistence of human papillomavirus as a
In addition, because we know that the shorter the excised ca- predictor for treatment failure after loop electrosurgical excision procedure.
nal length, the better the obstetric future, we believe that with this Int J Gynecol Cancer 2007;17:1271–7.

© 2022, ASCCP 5

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.

Foggiatto et al. Journal of Lower Genital Tract Disease • Volume 27, Number 1, January 2023

16. Costa S, De Simone P, Venturoli S, et al. Factors predicting human 21. Gatta LA, Kuller JÁ, Rhee EHJ. Pregnancy outcomes following cervical
papillomavirus clearance in cervical intraepithelial neoplasia lesions treated conization or loop electrosurgical excision procedures. Obstet Gynecol
by conization. Gynecol Oncol 2003;90:358–65. Surv 2017;72:494–9.
17. Gardeil F, et al. Persistent intraepithelial neoplasia after excision for cervical 22. Kyrgiou M, Athanasiou A, Paraskevaidi M, et al. Adverse obstetric
intraepithelial neoplasia grade III. Obstet Gynecol 1997;89:419–22. outcomes after local treatment for cervical preinvasive and early invasive
18. Lin C, Tseng CJ, Lai CH, et al. Value of human papillomavirus deoxyribonucleic disease according to cone depth: systematic review and meta-analysis. BMJ
acid testing after conization in the prediction of residual disease in the subsequent 2016;28:354–i3633.
hysterectomy specimen. Am J Obstet Gynecol 2001;184:940–5. 23. Baldauf JJ, Baulon E, Thoma V, et al. Obstetric outcomes LOOP-excision.
19. Mitchell MF, Tortolero-Luna G, Cook E, et al. A randomized clinical trial of J Gynecol Obstet Biol Reprod (Paris) 2013;42:534–40.
cryotherapy, laser vaporization, and loop electrosurgical excision for treatment of 24. Prendiville W, Cullimore J, Norman S. Large loop excision of
squamous intraepithelial lesions of the cervix. Obstet Gynecol 1998;92:737–44. transformation zone (LLETZ). A new method of management for
20. Nuovo J, Melnikow J, Willan AR, et al. Treatment outcomes for squamous women with cervical intraepithelial neoplasia. Br J Obstet Gynaecol
intraepithelial lesions. Int J Gynaecol Obstet 2000;68:25–33. 1989;96:1054–60.

6 © 2022, ASCCP

Copyright © 2022 ASCCP. Unauthorized reproduction of this article is prohibited.