Professional Documents
Culture Documents
org
GYNECOLOGY
Recurrence of high-grade cervical abnormalities
following conservative management of cervical
intraepithelial neoplasia grade 2
Tom M. Wilkinson, MBChB; Peter H. H. Sykes, MBChB, FRANZCOG;
Bryony Simcock, MD; Simone Petrich, MD
OBJECTIVE: Conservative management of cervical intraepithelial RESULTS: A total of 683 women were included: 106 with CIN 2 that
neoplasia (CIN) grade 2 in women younger than 25 years may reduce spontaneously regressed, 299 with treated CIN 2, and 278 with
overtreatment. However, long-term efficacy remains uncertain. This conservatively managed CIN 1. Median follow-up was 4 years. There was
retrospective cohort study aimed to determine the rate of recurrence of no significant difference in the risk of development of high-grade ab-
high-grade abnormalities among young women with a history of CIN 2 normalities after 2 years between the spontaneously regressing CIN 2
that spontaneously regressed within 2 years and compare this with the and CIN 1 groups (P ¼ .83). Women with treated CIN 2 had a significantly
rate of high-grade abnormality in similar women with an initial diag- lower risk of recurrence than women with untreated CIN 2 (P ¼ .01).
nosis of CIN 1.
CONCLUSION: CIN 2 that has spontaneously regressed appears to
STUDY DESIGN: We identified all women aged younger than 25 years behave as a low-grade lesion. This study contributes to the growing
who were diagnosed with CIN 1 or CIN 2 between January 2005 and body of evidence that careful observation of CIN 2 is an efficacious and
August 2009 within 2 colposcopy units. Follow-up data from the appropriate initial management option for women aged younger than
National Cervical Screening Programme were obtained to identify 25 years at diagnosis.
those women who developed recurrent high-grade lesions before
Key words: cervical intraepithelial neoplasia, conservative manage-
October 2012. Comparisons were made using Cox proportional haz-
ment, grade 2, young women
ards regression.
Cite this article as: Wilkinson TM, Sykes PHH, Simcock B, et al. Recurrence of high-grade cervical abnormalities following conservative management of cervical
intraepithelial neoplasia grade 2. Am J Obstet Gynecol 2015;212:769.e1-7.
than CIN 2,9 this risk is likely to be an of the cervix among women with a his- All screening records were compared
overestimate when considering CIN 2 tory of spontaneously regressing CIN 2 against inclusion and exclusion criteria
alone. Furthermore, it has been shown and compared this with the rate of high- for the cohort. Women were excluded if
that 40-74% of CIN 2 in women younger grade abnormality in similar women they had fewer than 2 years of available
than 25 years spontaneously regresses with an initial diagnosis of CIN 1. follow-up data, if they opted out of the
within 24 months.13-18 As a secondary aim, we also intended screening register, or if they received
In light of this fact and the harms of to compare with recurrence in women treatment for an initial diagnosis of CIN
intervention (which may include ob- who have received treatment for CIN 2. 1. Women were also excluded if both their
stetric complications19-22), international last result prior to 24 months and their
guidelines have shifted to recommend first result after 24 months indicated a
considering conservative management M ATERIALS AND M ETHODS high-grade abnormality, with no docu-
in young women with histologically Our study cohort was derived from 2 mented treatment between these 2 results
proven CIN 2.1,23,24 The 2006 American colposcopy units in New Zealand (ie, if there was no documented treatment
Society of Colposcopy and Cervical Pa- (Christchurch and Dunedin). or regression) because this would violate
thology consensus guidelines suggest a 6 Data were extracted retrospectively; both conventional treatment and con-
month observation with colposcopy and therefore, the study investigators had no servative management protocols. In
cytology for 24 months for HSIL, fol- influence on management decisions practice this means that all women in the
lowed by excision if the lesion is still made regarding cohort members. All study cohort were free of known high-
present.24 Perkins et al25 in 2012 found cohort members received a new histo- grade disease at 2 years of follow-up.
that the introduction of this guideline logical diagnosis of CIN 1 or CIN 2 be- Women with an initial diagnosis of
reduced the use of loop electrosurgical tween January 2005 and August 2009. At CIN 2 were divided into 2 groups:
excision procedures to treat moderate this time, local management guidelines spontaneous regression and treatment.
dysplasia in women aged 18-23 years were in favor of a conservative approach The spontaneous regression group
from 55% to 18%. The 2012 American to managing CIN 1, consisting of included all women in whom there was
Society of Colposcopy and Cervical Pa- 12 month cytological surveillance for 24 evidence of regression within 2 years
thology guidelines also agree with 24 months and in favor of immediate treat- of initial diagnosis and no evidence of
months of follow-up.26 ment for CIN 2.1 treatment. The treatment group included
Prior research has shown no progres- A selection of cohort members with all women who received some form of
sion to cancer among conservatively a diagnosis of CIN 2 were managed treatment for cervical dysplasia within
managed women with CIN 2 and a high conservatively with 6 month cytological 2 years.
rate of regression in both the adolescent and colposcopic surveillance for 24 Evidence of regression was defined as
and 20-25 year old age groups.14-17 How- months; however, this decision was at least 1 histology or cytology result
ever, these studies have encapsulated only dependent on individual patient and showing normal cervix or low-grade
short-term outcomes (up to 24 months of clinician preferences. changes only (CIN 1, HPV infection
follow-up). There is therefore an unex- Local colposcopy records at each unit only, low-grade squamous intraepithelial
plored possibility that, following apparent were searched to identify all women who lesion, atypical squamous cells of unde-
spontaneous regression, women with received a new histological diagnosis of termined significance), in the absence of
conservatively managed CIN2 may be at CIN 1 or CIN 2 between January 2005 a second sample taken concomitantly
increased risk of recurrent high-grade ab- and August 2009 and who were aged showing high-grade changes (CIN2, CIN
normalities. This would suggest dimin- younger than 25 years at the time of 3, HSIL, atypical squamous cells: high
ished efficacy of conservative management diagnosis. These records were also used grade squamous intraepithelial lesion
(with treatment being delayed rather than to determine age at diagnosis, ethnicity, cannot be excluded [ASC-H], adenocar-
prevented), create a need for close obser- smoking status at diagnosis, parity at cinoma in situ, and atypical glandular
vation beyond 2 years, and raise questions diagnosis, and to determine whether any cells).
of safety. Investigation is therefore woman had undergone hysterectomy. Treatments included local loop exci-
necessitated. Full cervical screening records for sion of the transition zone, cold knife
CIN 1 is generally accepted to be a these women were obtained from the cone biopsy, and laser ablation.
manifestation of human papillomavirus National Cervical Screening Programme In accordance with our cohort selec-
(HPV) infection only, carrying a very register. This provided records of all tion criteria, all included women with
low risk for subsequent malig- smear tests and colposcopies received by CIN 1 did not receive any treatment
nancy.9,27,28 This group therefore pro- the women, across all centers in New within 2 years of initial diagnosis so were
vides a good measure of what an Zealand, in both the public and private comparable in management with the
acceptable degree of recurrence is. health care sector, along with cytological CIN 2 spontaneous regression group.
This retrospective cohort study aimed and histological results, and a record We anticipated the total number of
to determine the long-term rate of of any treatments received, prior to women with CIN 1 being too large to be
recurrence of high-grade abnormalities October 2012. logistically manageable. Accordingly, we
TABLE 1
Descriptive data for the cohort
CIN 1 (conservative CIN 2 spontaneous CIN 2
Variable management) regression treatment Sig (P value)a
Age at diagnosis, y
<20 46 (18%) 32 (31%) 45 (15%) .002
20 214 (82%) 72 (69%) 253 (85%)
Age range, y (mean) 16e24 (21.4) 16e24 (20.7) 17e24 (21.7) < .001
Nulliparous at diagnosis
Yes 194 (83%) 67 (80%) 189 (76%) .16
No 40 (17%) 17 (20%) 60 (24%)
Hysterectomy during follow-up
Yes 0 (0%) 0 (0%) 0 (0%) N/A
No 178 (100%) 75 (100%) 165 (100%)
Smoker at diagnosis
Yes 79 (42%) 31 (36%) 82 (46%) .37
No 110 (58%) 54 (64%) 98 (54%)
Ethnicity
NZ European 200 (81%) 84 (88%) 235 (82%) .07
NZ Maori 19 (8%) 8 (8%) 27 (9%)
Pacific Islander 1 (0.4%) 0 (0%) 6 (2%)
Other 28 (11%) 4 (4%) 18 (6%)
Average gap between 1.07 1.02 1.20 .020
follow-up visits, y
Mean duration of 4.08 (2.00e7.36) 3.96 (2.02e6.92) 4.55 (2.03e7.73) < .001
follow-up, y (range)
Median duration of 3.86 3.69 4.45
follow-up, y
Five year follow-up 58 (21%) 20 (19%) 97 (32%) .001
data available
Year of initial diagnosis
2005 40 (14%) 12 (11%) 63 (21%) < .001
2006 30 (11%) 16 (15%) 69 (23%)
2007 78 (28%) 23 (22%) 76 (25%)
2008 83 (30%) 29 (27%) 61 (20%)
2009 47 (17%) 26 (25%) 30 (10%)
CIN, cervical intraepithelial neoplasia; N/A, not available; NZ, New Zealand; Sig, significance.
a
Determined from the X2 distribution for categorical variables, and using one-way analysis of variance for continuous variables.
Wilkinson. Recurrence of CIN2 in women managed conservatively. Am J Obstet Gynecol 2015.
with regard to percentage of women between these 2 groups in frequency of abnormalities (calculated from Cox
with 5 year follow-up data (P ¼ .76), follow-up (P ¼ .54). proportional hazards regression) are
in average length of follow-up (P ¼ Kaplan-Meier curves for the 3 groups presented in Table 2. Adjustments were
.43), or year of initial diagnosis (P ¼ are presented in Figure 2. Hazard ratios made for age at diagnosis, smoking
.25). There was also no difference for development of high-grade status at diagnosis, ethnicity, and initial
TABLE 2
Hazard ratios for development of high-grade abnormalities after 2 years
Hazard ratio (raw) Hazard ratio (adjusted)
Comparison (95% CI) Sig (P value) (95% CI)a Sig (P value)
CIN 2 conservative compared with CIN 1 0.87 (0.43e1.77) .70 0.92 (0.44e1.94) .83
CIN 2 conservative compared with 2.98 (1.30e6.86) .01 2.97 (1.24e7.09) .01
CIN 2 treatment
CIN 1 compared with CIN 2 treatment 3.44 (1.80e6.58) < .001 3.23 (1.66e6.29) .001
CI, confidence interval; CIN, cervical intraepithelial neoplasia; Sig, significance.
a
Adjusted for smoking status, age, ethnicity, and treatment center.
Wilkinson. Recurrence of CIN2 in women managed conservatively. Am J Obstet Gynecol 2015.
the initial diagnosis onward for women The percentage of women in each group one would expect that it may result in a
from each center was calculated. This who developed high-grade abnorma- higher rate of high grades in this group,
was higher for Christchurch (0.41) than lities is apparently high: 17% for the CIN so exclusion of a subgroup is unlikely to
Dunedin (0.33), although the difference 2 spontaneous regression group, 12% alter the conclusion of the study.
was not significant (P ¼ .14). for the CIN 1 group, and 4% for the CIN Any bias introduced by varied follow-
2 treatment group. However, this is up regimens is minimized by the use of
unsurprising, given the study design, data from a national screening program
C OMMENT which purposely took a very generous (allowing for systematic identification of
To the best of our knowledge, this is the approach, requiring only 1 result (cytology patients and ensuring a degree of regular
first time long-term outcomes (beyond 2 or histology), suggesting a high-grade follow-up), and the impact of differing
years) of CIN 2 conservative manage- change to declare that such an abnormal- management philosophies at different
ment have been reported in the litera- ity was present. This ensured sufficient centers is apparent. Regardless, the main
ture. The main finding of this study is power and was applied consistently be- findings of the study remained the same
that, where CIN 2 is present in a young tween groups. However, it does mean after adjusting for treatment center,
woman (age younger than 25 years) and that the apparent rate of such abnor- indicating a good degree of generaliz-
spontaneously regresses within 2 years, malities reported is likely to be an ability. Furthermore, the observed dif-
the risk of recurrent high-grade abnor- overestimate. ferences between centers provide insight
malities is not significantly higher than The effect of this approach can be best into the difficulties of interpretation of
the risk among women initially diag- seen in the difference noted between the the studies and management protocols
nosed with CIN 1. This affirms the value 2 study centers, with women in Christ- reliant on a histological and cytological
of conservative management of CIN 2 in church having a higher risk of high- diagnosis of CIN grade.
this group by demonstrating that a large grade abnormality than women in It should also be acknowledged that
subset of CIN 2 behaves as a low-grade Dunedin. Post-hoc analysis suggested there is clear evidence of inter- and
lesion (and hence a case can be made that if only cytological results were intraobserver variation in histological
for managing it as one). considered, there would be no difference and cytological diagnosis of cervical ab-
Specific inclusion and exclusion in the observed rate of high-grade ab- normalities.29-31 In this study, results
criteria mean that these 2 groups of normality between the 2 centers. were not adjudicated, and therefore,
interest (CIN 1 and CIN 2 conservative) This is likely to be in part because of there may be inaccuracies. In particular,
were comparable within the study more biopsies being carried out per results initially thought to be CIN 2 may
cohort. No differences were found in the woman in Christchurch (reflected in the actually represent CIN 1. Although this
distribution of year of initial diagnosis, greater number of histology results per is an important caveat, it is not a signif-
frequency of follow-up, or length of woman and in follow-up being slightly icant limitation, given the clinical
follow-up, so the results are unlikely to more frequent on average), rather than a context of the question at hand. When
be biased by changing management higher rate of cytological abnormalities. deciding on an appropriate management
protocols or differential follow-up. Clearly, differences in clinical approaches plan, the clinician must act on the basis
Although the 2 groups did signifi- to colposcopy can influence the observed of the data available, and therefore, our
cantly differ in age at diagnosis (with rate of cervical dysplasia. Reassuringly, study groups were also classified on this
CIN 2 women more likely to be younger controlling for the difference between same basis.
than 20 years), adjusting for this in the the 2 study centers did not substantially A further implication of the variability
analysis did not alter the result. alter the study findings (the raw and of testing is that CIN 2 may in fact cover
Furthermore, all women in both groups adjusted hazard ratios were similar). a range of biological entities. Our main
underwent a biopsy for the purposes of This illustrates an important weakness group of interest was women who had
providing a histological diagnosis. Any in the study design: its retrospective na- CIN 2 spontaneously regress within 2
possibility of such a biopsy having an ture means that patient management years, and this is likely to represent a
impact on the natural history of CIN 1 or and follow-up was not standardized. A specific subset of all CIN 2. In particular,
2 has therefore been negated for the liberal definition of regression of CIN 2 it may represent a subset with an un-
purposes of comparison. was applied, including all women with at derlying biology similar to CIN 1.
In comparison with women who have least 1 cytology or histology specimen Further research is needed into the
undergone treatment, conservatively confirming CIN 1 or less and no evi- stratification of CIN 2, in particular
managed women appear to remain at an dence of subsequent high-grade change looking for predictors of regression.
increased risk of high-grade abnormality within 24 months. Although we would Because of the small size and retro-
for at least 5 years. This finding has im- normally expect 2 such specimens spective nature of this study, larger pro-
plications for conservative management because of the heterogeneous nature of spective studies need to be performed
protocols, and full exploration of this this group and allowing for delays in before definitive conclusions can be
effect requires follow-up on a longer follow-up, a stricter definition could not made. The cohort size was determined
time scale than that present here. be applied. If this would introduce a bias, by the number of women who happened