SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

Objective To analyse biopsies of large loop excision of the transformation zone of the cervix; to identify factors associated with negative histology; and to develop predictive models in order to reduce the number of negative loop excisions.

Design Retrospective analysis of patient notes and audit database.

Setting Colposcopy clinic of a large district general hospital in North Staffordshire.

Population Four hundred and fifty-two women who underwent a large loop excision of the transformation zone (LLETZ) procedure for suspected cervical intraepithelial neoplasia.

Methods Women who underwent a LLETZ procedure were placed in two different groups, one positive for cervical intra epithelial neoplasia and the other negative for cervical intra epithelial neoplasia. Information was obtained on a number of clinical and colposcopic variables. Analysis was undertaken to determine if there were any differences between the two groups. These factors were then identified and three predictive models generated. Receiver-operator characteristic curves were used to assess and test these models.

Main outcomes measures To identify factors associated with negative histology on a LLETZ specimen. To predict how to reduce the number of negative LLETZ specimens.

Results Four hundred and fifty-two women underwent a LLETZ procedure, 88 were negative (19%) and 364 were positive (81%). In women who were treated at their first visit, 56/316 (18%) had negative histology. There were significant associations between negative histology in the LLETZ and negative or low grade cytological atypia, negative colposcopic findings and years of age > 50 in both bivariate analysis and stepwise logistic regression. In the predictive models, the sensitivity ranged between 72% and 80%, the specificity 59%–72%, and the area under the receiver-operator characteristic was 0.75–0.77. If we had used the predictor models and managed women with negative or low grade cervical atypia and negative colposcopy findings conservatively, we would have reduced the negative biopsy rate from 19% to 14%, but five cases of high grade disease and 25 cases of low grade disease would have been missed. If we had also included women aged > 50 years in this model, the negative biopsy rate would have dropped from 19% to 15%, with only one case of high grade disease and 11 cases of low grade disease missed. All these women would require continued cytological and colposcopic surveillance. Importantly, no cases of invasion would have been missed.

Conclusion Using a predictive model can reduce the number of negative LLETZ specimens, but at the expense of continued cytological and colposcopic surveillance and cannot be recommended in normal practice. This raises the question whether current standards for negative histology in LLETZ specimens are set unrealistically high.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The cervical screening programme is achieving its goal of reducing deaths from cervical cancer, but at what cost? Far more women will undergo investigation and treatment for suspected cervical intraepithelial neoplasia than are likely to develop cervical cancer. Consequently more women may be harmed than saved. Large loop diathermy excision of the cervical transformation zone (LLETZ) is widely used and is now one of the commonest mode of investigation and treatment for cervical intra epithelial neoplasia1.

Ease of use has significantly contributed to its popularity, but with this come the dangers of overtreatment. Concern has been expressed about the proportion of LLETZ biopsies that contain no cervical intra epithelial neoplasia and how this might be influenced by treatment strategies, such as see-and-treat2. In previous studies negative biopsy rates of between 4.7% and 41.0% have been reported3,4. To address these anxieties, the NHS cervical screening programme requires that > 85% of all biopsies should contain cervical intra epithelial neoplasia2, and this should be > 90% for patients treated at their first consultation5. However a recent report indicates that most colposcopy clinics fail these standards1.

We have regularly audited our colposcopic practise since 1990, during which time our treatment strategies changed from an interventionist see-and-treat policy (LLETZ rate of 56% overall and 82% in new patients) to a highly selective policy (33% and 22%, respectively). Although there was an increase in the proportion of LLETZ specimens containing cervical intra epithelial neoplasia III, there was also a significant fall in the actual amount of cervical intra epithelial neoplasia III diagnosed as this strategy lacked sensitivity. Accordingly, our LLETZ selection criteria were made less stringent. Despite this, the proportion of negative LLETZ specimens has consistently remained above 15% (19–23%). In order to reduce the number of negative biopsies, we undertook an analysis of LLETZ biopsies to see if negative histology can be predicted.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The patient sample comprised all women who underwent a LLETZ procedure performed in a single consultant's (C.W.E.R.) colposcopy clinic between 1 January 1998 to the 31 December 1998. Information obtained regarding the patient included; age, referral pattern (e.g. new patient, follow up patient or re-referral), previous treatment, number of previous pregnancies, smoking history (e.g. smoker, non-smoker or previous smoker), presence and method of contraception and menopausal status.

Colposcopy was performed by one of three British Society for Colposcopy and Cervical Pathology certified colposcopists (n= 343), or one of two trainees under direct supervision (n= 109). Information obtained at the time of colposcopic examination included the name of the colposcopist, the referral smear, colposcopic findings and whether or not the squamo-columnar junction could be visualised. The colposcopist was aware of the initial cytology report at referral, and repeat cytology was not routinely undertaken at our centre before performing an excisional procedure. LLETZ was performed without prior histological confirmation of atypia, except for three cases who had cervical intra epithelial neoplasia diagnosed by punch biopsy. During the period under review the indications for LLETZ were: moderate or severe dyskaryosis; colposcopic evidence of epithelial atypia; dyskaryosis and whole of cervical transformation zone not visible; and persistent dyskaryosis after previous treatment or 24 months observation.

Lastly, the histological result was obtained. For the purposes of this analysis the patient sample was classified according to the presence or absence of cervical intra epithelial neoplasia into two groups: Group one (positive for cervical intra epithelial neoplasia) and Group two (negative for cervical intra epithelial neoplasia). Negative histology was defined as the absence of cervical intra epithelial neoplasia and positive histology was defined as the presence of cervical intra epithelial neoplasia (including glandular lesions and basal layer atypia) or worse.

The size of the loop biopsy varied, but as far as we are aware there were no cases reported of inadequate sampling of the transformation zone in patients with negative histology. The clinical and colposcopic variables were collected by individuals masked to the histology but the individuals reporting the histology were not masked to the clinical and colposcopic information.

The main end point of the study was to identify factors (both clinical and colposcopic) associated with negative histology on a LLETZ specimen. From this data, we would be able to develop a predictive model in order to reduce the number of negative LLETZ specimens.

Statistical analysis

Statistical analysis was undertaken using the software package Stata Version 5.0 (College Station, Texas, USA). The association between a number of variables and the histological results were studied using an unpaired Student's t test for continuous data, when normally distributed and the χ2 test for categorical data.

In addition, a bivariate (corrected for age) and multivariate logistic regression analysis was undertaken. Most predictors were binary, in the case of referral pattern and age, these were transformed to binary variables. Categorical data were expressed in data sets of binary variables with a reference category. As we were trying to predict a negative result, we coded positive histology for cervical intra epithelial neoplasia as 0 and negative histology as 1.

Reference categories were: age < 50 years versus > 50 years; new referrals versus follow up visits; no previous treatment versus previous treatment; nulliparity versus parity; nonsmoker versus smoker; no contraception versus presence of contraception; postmenopausal versus premenopausal status; negative smear or low grade atypia versus moderate or high grade atypia; negative colposcopic findings versus positive findings; and nonvisualisation of the squamo-columnar junction versus visualisation of the squamo-columnar junction. A P value of ≤ 0.05 was considered significant.

As many of these variables were necessarily interrelated, a stepwise logistic regression analysis (using a step-up model) was performed to identify those variables that have the greatest individual association with histological outcome and which together form the best predictive model for negative histology. For the step-wise model, a value of P < 0.1 was used for the inclusion of a predictor in the model.

Finally, three predictor models (see Table 4 later) were generated from the logistic regression. In the first predictive model (Model 1), all the clinical and colposcopic variables were used. In the second predictive model (Model 2), only the significant variables determined by the stepwise analysis were used. In the third predictive model (Model 3), only the significant variables were used as mentioned above and what the authors felt were the most important clinical variables in colposcopic assessment. A receiver-operator characteristic curve was then used to assess and compare the three different models. The receiver-operator characteristic curve was also used to determine the cut off for the predictive probability of negative histology, which yielded values for sensitivity, specificity and positive predictive value. The area under the receiver-operator characteristic curve was used to measure the diagnostic accuracy of the three models6.

Table 4.  The three predictor models for predicting negative histology in a large loop excision of the transformation zone (LLETZ) specimen. SCJ = squamo-columnar junction.
 Code for modelRegression coefficientStandard errorProbability level
Model 1    
 Intercept −1.080.620.08
 Age in years> 50 years = 0, < 50 years = 10.860.610.16
 Referral patternFirst visit = 0, follow up = 10.0860.310.78
 Previous treatmentNone = 0, Yes = 10.690.380.07
 ParityNo = 0, Yes = 10.510.390.19
 Ex-smokerNo = 1, Yes = 10.350.350.32
 SmokerNo = 0, Yes = 1−0.630.350.08
 ContraceptionNo = 0, Yes = 10.0380.360.91
 Menopausal statusPre = 0, Post = 10.0850.610.89
 Referral smear (grade)Low/neg = 0, high = 1−1.060.28< 0.001
 Colposcopic findingsNegative = 0, positive = 1−1.280.31< 0.001
 SCJ seenNo = 0, Yes = 10.350.340.31
Model 2    
 Intercept−0.260.270.35 
 Age in years> 50 years = 0, < 50 years = 10.930.300.002
 Referral smearLow/neg = 0, high = 1−1.070.27< 0.001
 Colposcopic findingsNegative = 0, positive = 1−1.280.28< 0.001
Model 3    
 Intercept−0.530.340.12 
 Age > 50 years> 50 years = 0, < 50 years = 11.020.320.001
 Referral smearLow/neg = 0, high = 1−1.110.27< 0.001
 Colposcopic findingsNegative = 0, positive = 1−1.290.30< 0.001
 SCJ visualisedNo = 0, Yes = 10.250.330.45
 Previous treatmentNone = 0, Yes = 10.630.350.07

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

During the study period, 1149 patients were seen (new = 724; follow up = 255; re-referral = 170). The mean age of clinic attenders was 37.5 years (SD 12.03; range 18–74 years). Twenty-three patients were not included in the analysis, because they underwent treatment other than LLETZ (10 underwent cone biopsy and 13 hysterectomy). In total, 452 patients underwent a LLETZ procedure (new = 316, follow up = 65; re-referral = 71), including three patients who underwent LLETZ after an earlier punch biopsy had demonstrated cervical intra-epithelial neoplasia. The treatment rates were 40% (n= 452) for the clinic as a whole, and 44% (n= 316) for new patients when this was performed at or as a result of findings at the first visit.

One hundred and seventy-five women (39%) underwent a LLETZ for colposcopic findings, 232 (51%) because of the referral smear and 45 (10%) because of incomplete examination or to remove a polyp or because of persistent dyskaryosis. Overall, 88 (19%) of the LLETZ samples were negative and 364 were positive (81%), with 268 (59%) containing cervical intra-epithelial neoplasia II or worse (high grade disease). In women who were treated at their first visit, 56 (18%) had negative histology. The mean age of patients with negative histology was 44.4 years compared with the mean age of patients with positive histology of 35.8 years (P < 0.001).

The patient characteristics of the two study groups are summarised in Table 1 and the colposcopic findings are summarised in Table 2. Negative biopsy was associated with increasing age of women, previous treatment to the cervix, parity, smoking, use of contraception, menopausal status, negative or low grade referral smear, low grade colposcopic findings and nonvisualisation of squamocolumnar junction. There was no association between who performed the LLETZ and negative histological result. It is interesting to note that only seven women (5%) out of 127 women with severely dyskaryotic smears had negative LLETZ histology.

Table 1.  Comparison of the patient characteristics between the two study groups (ntotal= 452). Values are given as median [SD] {95% CI} or n (%), unless otherwise indicated.
 Positive histology (Group 1) (n= 364)Negative histology (Group 2) (n= 88)P
  1. *Unpaired Student's t test.

  2. χ test.

Age35.83 [0.60] {34.65–37.01}44.39 [1.28] {41.84–49.93}< 0.001*
Referral pattern   
 First visit260 (57)56 (12)0.10
 Follow up visit104 (23)32 (8) 
Previous treatment   
 No328 (72)71 (16)0.01
 Yes35 (7)17 (4) 
 Missing data1 (<1)0 (0) 
Parity   
 No82 (18)10 (2)0.02
 Yes282 (62)78 (18) 
Smoker   
 No183 (40)59 (14)0.005
 Yes181 (40)29 (6) 
Ex-smoker   
 No125 (28)36 (7)0.25
 Yes239 (53)52 (12) 
Contraception   
 No110 (24)39 (9)0.01
 Yes254 (56)49 (11) 
Menopausal status   
 Premenopausal315 (69)60 (13)<0–001
 Postmenopausal48 (10)28 (7) 
 Missing data1 (<1)0 (0) 
Table 2.  Comparison of the colposcopic characteristics between the two study groups (n= 452). Values are given as n (%), unless otherwise indicated. SCJ = squamo-columnar junction; ESI = early stromal invasion.
 Positive histology (Group 1) (n= 364)Negative histology (Group 2) (n= 88)P
  1. *χ2 test.

  2. Includes 6 smears: (scanty smears, inappropriate endometrial cells, postmenopausal smear).

Referral smear   
 Negative8 (2)10 (2)< 0.001*
 Borderline/mild dyskaryosis90 (20)40 (9) 
 Moderate dyskaryosis127 (28)16 (4) 
 Severe dyskaryosis120 (26)7 (1) 
 Glandular17 (4)11 (2) 
 Others2 (<1)4 (1) 
Colposcopic findings   
 Negative57 (13)41 (8)< 0.001*
 Low grade (CIN I)70 (15)15 (3) 
 High grade (CIN II/III)212 (49)14 (3) 
 Uncertain19 (4)18 (4) 
 Invasion6 (1)0 (0) 
SCJ seen   
 No69 (15)32 (7)< 0.001*
 Yes294 (65)56 (12) 
 Missing data1 (<1)0 (0) 
Histology   
Negative 88 (19) 
CIN I or low grade CGIN (Low grade disease)97 (21)  
CIN II/III or high grade CGIN (High grade disease)260 (58)  
ESI/invasion7 (2)  

In the bivariate logistic regression, however, only the presence of a negative or low grade referral smear and negative colposcopic findings were significantly associated with negative histology (Table 3). Interestingly, previous treatment and visualisation of the squamo-columnar junction were not associated with negative histology. In the stepwise analysis, only three variables were involved in the model for predicting negative histology. These were low grade cytological atypia (i.e. mild dyskaryosis or less) (P= 0.001); negative colposcopic findings (P < 0.001); and if the woman was aged > 50 years (P= 0.03).

Table 3.  Bivariate logistic regression analysis of patient and colposcopic variables (corrected for actual age). BNA = borderline nuclear atypia; SCJ = squamo-columnar junction.
 POdds ratio95% CI
  1. The reference category for each variable was given an odds ratio of 1.0.

Referral pattern   
 First visit 1.00 
 Follow up visit0.141.480-.88–2.47
Previous treatment   
 No 1.00 
 Yes0.071.820.94–3.52
Parity   
 No 1.00 
 Yes0.481.310.62–2.77
Smoker   
 No 1.00 
 Yes0.110.660.39–1.10
Ex-smoker   
 No 1.00 
 Yes0.710.910.55–1.50
Contraception   
 No 1.00 
 Yes0.501.230.67–2.25
Menopausal status   
 Pre 1.00 
 Post0.390.690.30–1.60
Referral smear   
 Negative 1.00 
 Low grade (BNA/mild dyskaryosis)0.0182.811.20–6.62
 High grade (Moderate/severe dyskaryosis)< 0.00110.063.90–25.98
 Glandular0.801.160.37–3.57
Colposcopic findings   
 Negative 1.00 
 Low grade (CIN I)0.0030.350.17–0.71
 High grade (CIN II or greater)< 0.0010.130.07–28
SCJ seen   
 No 1.00 
 Yes0.710.890.49–1.63

For all the models we estimated the probability in order to determine the cut off point. In Model 1 (which contained all the clinical and colposcopic variables) (Table 4), using a cut off of 0.18 for the predictive probability, the sensitivity was 72%, the specificity 72% and positive predictive value 37%. The area under the curve for the first predictive model was 0.77 (SD 0.032). In Model 2 (containing the variables; smear grade, colposcopic findings, age category (older or younger than 50 years) (Table 4), at a cut off of 0.18 for the predictive probability, the sensitivity was 80%, the specificity 57% and the positive predictive value 30%. The area under the curve for the second predictive model was 0.75 (SD 0.032).

In Model 3 this contained the three significant variables mentioned above plus visualisation or nonvisualisation of the squamo-columnar junction and whether or not the patient had undergone previous treatment (Table 4). We felt whether the squamo-columnar junction was visualised and whether the woman had undergone previous treatment were important determinants on electing to perform a LLETZ if the colposcopic examination was incomplete. For this model, at a cut off of 0.18 for the predictive probability, the sensitivity was 79%, the specificity was 59% and the positive predictive value 30%. The area under the curve for the third predictive model was 0.75 (SD 0.032).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The objective of this study was to identify what factors could be used to predict the absence of cervical intraepithelial neoplasia in a specimen following LLETZ of the cervix. LLETZ was introduced approximately 10 years ago for the treatment of pre-malignant conditions of the cervix7. It appears to be the most popular method for excisional procedures accounting for 44% of all treatments performed1, and for 59% of treatments at first visits1. Negative histology, however, is not unique to LLETZ, with rates between 20% and 34% being reported for laser cone biopsies8,9, and of 41% for punch biopsy10.

Despite recommendations that nearly all LLETZ specimens should contain cervical intra epithelial neoplasia, this has not been achieved in the majority of colposcopy units throughout the country1. We did not achieve the standard recommended by the National Health Service Cervical Screening Programme either in the patient group as a whole or in women treated at their first visit, with a negative histology rate of 19% (for all loops) and 18% (for first visits). We felt, that if factors that determine negative histology could be identified, it may be possible to reduce the number of negative loops so that national standards could be realistically achieved.

In this study, we found that in the logistic regression analysis negative histology was strongly associated with older age, negative or low grade referral cytology and negative colposcopic findings. It has been previously reported that there was no association between negative histology (from a laser cone) and age9, but this contrasts with our findings where the mean age of those women with a negative LLETZ specimen was significantly different from the mean age of women with positive histology. This may, however, reflect the change in practice in screening and cytology over the last 10 years.

An alternative explanation could be that due to the position of the transformation zone in women aged > 50 years being in the endocervical canal, it is less likely to be removed adequately or sampled using LLETZ. Certainly, the squamo-columnar junction is seen far less at colposcopy in this group of women. We would argue, however, that the transformation zone was adequately sampled in our clinic, because we usually perform a loop cone or two-stage LLETZ in this group of women.

We found that the likelihood of a LLETZ specimen being negative was increased the lower the grade of the initial referral smear, which agrees with Denny et al.11 who found that in cases of negative LLETZ, 44% had a initial low grade smear. Lastly, negative colposcopic findings were associated with a negative LLETZ specimen. It is well established that colposcopy is not very accurate. Indeed, Skehan et al.8 previously reported the colposcopic accuracy of diagnosing cervical intra-epithelial neoplasia was approximately 60%. A recent meta-analysis has demonstrated that while colposcopy performs well in the context of high grade smears, it is less predictive in the context of low grade smear and when the colposcopy findings are low grade12.

When discussing the results and comparing with other studies, it must be stressed that we examined the factors associated with negative histology, whereas most other work has concentrated on predicting positive outcome. In this study, the sensitivity, specificity, predictive values and accuracy of the three models compare favourably with established methods for the detection of pre-malignant disease of the cervix. For Model 1, the sensitivity in predicting negative histology was 72%, with a specificity of 72% and the area under the receiver-operator characteristic was 0.77.

Skehan et al.8 found that the sensitivity for cervical cytology was 83% and for colposcopy 89% in determining positive histology with LLETZ. The specificity, however, was low at 14% and 17%, respectively, and the overall accuracy was 62% for smears and 65% for colposcopy. Previous studies, which have tested the effectiveness of colposcopy in distinguishing normal tissue against varying degrees of cervical intraepithelial neoplasia, have found that sensitivity (64–99%) 12–14 is better than specificity (30–93%)12,15,16, although there is wide variation.

There was no loss of sensitivity with Model 2 (containing only the significant clinical and colposcopic factors) compared with Model 1, but there was loss of specificity from 72% down to 57%. Loss of specificity in this model would lead to more women with positive histology being missed, which can be explained thus: In total, 64 women in the clinic presented with negative or low grade dyskaryosis, had a normal looking cervix at colposcopy and underwent a LLETZ. Of these 64 women, 34 women had negative histology and 30 positive histology. Had all of these women been managed conservatively, the overall treatment rate for the clinic would have dropped to 34% (388/1149) and the negative biopsy rate to 14% (54/388). For women treated at their first visit, the treatment rate would have dropped to 39% (281/724) and the negative biopsy rate to 14% (39/281).

Although 34 women would have been spared unnecessary intervention, 30 with disease would not have received treatment. There would have been five cases of high grade disease (cervical intra epithelial neoplasia II or worse) and 25 cases of low grade disease (cervical intra epithelial neoplasia I or less) missed using this model. This could be avoided by following up these women in the colposcopy clinic in six months, and if there was persistent cytological atypia a LLETZ could be performed at that stage. At present we are unable to determine whether the women who had negative histology had subsequent negative histology, but this would be an important point to address.

We also considered negative or low grade cytology, negative colposcopy findings and age > 50 years. Had all of these women been managed conservatively, the overall treatment rate for the clinic would have dropped to 36% (415/1149) and the negative biopsy rate to 15% (63/415). For women treated at their first visit, the treatment rate would have dropped to 41% (294/724) and the negative biopsy rate to 14% (42/294). There would have been one case of high grade disease, and 11 cases of low grade disease missed using this model. Importantly, using these models, no cases of invasive disease would have been missed.

In Model 3, we considered what was the most important colposcopic determinants of a woman undergoing an excisional procedure to the cervix. Satisfactory colposcopy entails that the entire limits of the colposcopic abnormality be defined otherwise the colposcopic assessment should be considered incomplete17. We felt that important factors that may influence this were whether the squamo-columnar junction was adequately visualised and whether the woman had undergone previous treatment. It is well known that assessment of the squamo-columnar junction is difficult in postmenopausal women. In order to facilitate complete colposcopic assessment in postmenopausal women, some would advise prior treatment to colposcopy with oestradiol18.

However, in Model 3, there was no loss of sensitivity—but a similar loss of specificity compared with the first model—, despite including these important colposcopic variables. This would imply that visualisation of the squamo-columnar junction and previous treatment should not influence whether a LLETZ will be negative and may not need to be considered when considering performing a LLETZ. It does therefore appear that age, referral cytology and colposcopic findings are significant determinants of negative histology at LLETZ and could be used in a predictive model to reduce the number of negative LLETZ specimens.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

Models can be used in order to predict which colposcopic and clinical variables may reduce the number of negative LLETZ procedures and thus prevent unnecessary treatment to the cervix. However, we cannot recommend its use in normal colposcopic practice because of the complexities of these models and because a small proportion of cases of cervical intra epithelial neoplasia would be missed using this technique, which requires future evaluation and assessment. A further conclusion would be that current standards for negative histology with LLETZ are targeted too high and we should accept more realistic figures.

Acknowledgements

We would like to thank all the nursing staff in the colposcopy suite for their assistance with this paper and to the colposcopy secretary Mrs S. Forrester for her help in collecting the data for this study.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References