METHODS

From December 1998 until March 2000, 75 women referred to the colposcopy clinic of the University Hospital Rotterdam (n = 63) and the University Hospital Vrije Universiteit in Amsterdam (n = 12) were asked to participate in our study. Four women with abnormal cervical cytology refused to participate. Of the 71 women enrolled in the study, 56 had abnormal cervical cytology (very mild dyskaryosis or worse) and 15 women had normal cervical cytology. The mean age of the participating women was 35 years (range, 20 to 63 years). After an explanation of the study and the use of the self sampling device by the study coordinator, written informed consent, approved by the ethics review boards of both participating hospitals, was obtained from each participant.

At intake, women received a cervicovaginal self sampling device, a form with detailed instructions, and a questionnaire on the use of the device. The self sampling device consisted of an irrigation syringe (50 ml; Bard Inc, Covington, UK), a disposable female urine catheter (single use female urine catheter ch.16; Astra Tech, Mölndal, Sweden), and a container with 15 ml sterile phosphate buffered saline (PBS) for irrigation. According to the instructions, the catheter had to be attached to the syringe to aspirate the irrigation fluid from the container. After aspiration, a cervicovaginal specimen was obtained by inserting the tip of the catheter as deep as possible into the vagina and pressing and releasing the balloon of the syringe three times, to flush the irrigation fluid into the vagina and back into the syringe. After removal of the catheter, the syringe, which contained the cervicovaginal specimen, had to be emptied in the container. Women were asked to obtain a cervicovaginal lavage the day before their return visit to the colposcopy clinic. At colposcopy, after introducing a vaginal speculum, the gynaecologist performed a cervicovaginal lavage with a similar device by irrigating the cervix and aspirating the fluid pooled in the posterior vaginal fornix. This was followed by a cervical smear obtained with a Cervex® brush (International Medical Products, Zutphen, the Netherlands). After a smear was made on to a glass slide the brush was placed in a buffer solution (PBS) and sent to the laboratory for HPV detection. Colposcopic examination followed and biopsy samples were taken for histological verification of suspected lesions. If necessary, women were treated according to a standard protocol. When no lesions were seen at colposcopy the cervix was considered to be free of disease (no cervical intraepithelial neoplasia (CIN)) and no biopsies were taken. The lavages, the cervical smear, and the brush for HPV detection were processed at the department of pathology at the Vrije Universiteit Medical Centre in Amsterdam. The lavages were vortexed and divided into two specimens. The first was used for cervical cytology reading, the second for HPV DNA testing.

Questionnaire

All participants were asked to fill in a questionnaire on the use of the self sampling device including the following questions: (1) What is your opinion about the use of the self sampling device? Answer: easy/difficult, and for what reason? (2) What screening tool would you prefer for your next screening round—self sampling or Pap smear? Answer: self sampling/Pap smear, and for what reason?

Cervical cytology

From each cytology specimen two cytospins were made and Pap stained. The cytology slides and biopsy samples were read by an expert pathologist who was unaware of the clinical findings. Cervical smears were classified according to the KOPAC classification, the standard classification in the Netherlands.¹⁶ This is a modification of the Pap classification.¹⁷ Cervical smears are cytomorphologically classified as Pap 1 (normal cytology), Pap 2 (very mild dyskaryosis), Pap 3a (mild to moderate dyskaryosis), Pap 3b (severe dyskaryosis), Pap 4 (suspected of carcinoma in situ), and Pap 5 (suspected of at least microinvasive carcinoma). Histology was classified as CIN0 (no dysplasia), CIN1 (mild dysplasia), CIN2 (moderate dysplasia), and CIN3 (severe dysplasia).

High risk HPV testing

The specimens for HPV testing were centrifuged at 2719 ×g for six minutes to pellet the cells. The supernatant was discarded and the pellet was suspended in 1 ml 0.01M Tris/HCl (pH 8.3) and stored at −80°C until further analysis. A β globin polymerase chain reaction (PCR) was performed to ascertain the quality of the target DNA. Testing for HPV was done by PCR enzyme immunoassay, which used HPV general primer mediated PCR with the general primers GP 5+/6+ to detect a broad spectrum of mucosotropic HPV types.^18,19 PCR products were used to identify in one assay all 14 high risk HPV types using enzyme immunoassay (types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68). This has been described previously and has been clinically validated.^19,20

Statistical analysis

The concordance between the cytology (Pap smear classification) and HPV results from the smear, from the lavage taken by the doctor (lavage-doctor), and from the lavage taken by the patient (lavage-patient) was calculated. The lavage-doctor was included in the study as a control for the lavage-patient, because the first was performed under optimal conditions; that is, the cervix was visible during irrigation. The κ value was computed as a measure of overall agreement beyond chance. A κ estimate of less than 0.2 indicates poor agreement, a κ estimate between 0.2 and 0.8 fair to moderate agreement, and a κ estimate of more than 0.80 good agreement.²¹ To compare the performances of HPV testing with cytology results in the smear and the lavages, we calculated for each test the sensitivity, specificity, and positive and negative predictive values to detect high grade CIN (CIN 2/3; moderate to severe dysplasia) and tested whether there were differences using the Mc-Nemar test. For each woman, the highest CIN grade in either the diagnostic cervical biopsy or the cervical tissue obtained at treatment was used as the reference for assessing test performance. Women with unsatisfactory cervical cytology—that is, no or too few cervical cells—or a β globin PCR negative test were not included in these analyses because we assumed that in a normal situation they would have been asked to repeat the self sampling or return for a repeat test.

RESULTS

Questionnaire on the use of the self sampling device

The use of the self sampling device was considered easy by 49 of the 56 (88%) women who performed a self sampling. The remaining seven women concluded that the device was difficult to use. They were uncertain about the amount of fluid they had aspirated and questioned the efficacy of the lavage. Three of them had normal cervical cytology.

At the next screening round, 13 of 56 (23%) of the women said that they would prefer the classic Pap smear to the self sampling. Their reasons were: (1) no problem with gynaecological examination (n = 8), and (2) the self sampling device is not practical (n = 5). The remaining group favoured the self sampling. All participating women regularly went to see their doctor for a Pap smear.

Fifteen women, five with normal and 10 with abnormal cervical cytology, did not perform the self sampling. Their reasons were: (1) they forgot to perform it (n = 10) and (2) they were too nervous about the colposcopic examination (n = 5). Three lavages by the doctor were unsatisfactory for cytological judgement. Four samples (lavage-doctor (n = 2) and lavage-patient (n = 2)) were β globin PCR negative.

Agreement of cytology and HPV testing between cervical smears, lavages, and self sampling specimens

There was fair agreement between the cytology results in the Pap smear and the lavage-doctor (κ = 0.28; table 1) with a 54% concordance (37 of 69 satisfactory slides). The agreement between the cytology results in the Pap smear and the lavage-patient was poor (κ = 0.14), with a concordance of 41% (23 of 56 slides). A fair agreement was obtained between the cytology results of the lavage-doctor and the lavage-patient (κ = 0.37; data not shown), with a 74% concordance between the cytology results performed on the two different samples.

The concordance between the high risk HPV test results in the smear and the lavage-doctor was 93% (64 of 69 women with β globin positive PCR tests), which is a good agreement (κ = 0.82; table 2). A 78% (39 of 48 women with β globin positive PCR tests) concordance between the HPV test results in the smear and those in the lavage-patient was seen, indicating moderate agreement (κ = 0.53). There was moderate agreement between the results obtained by HPV testing of the lavage-doctor and the lavage-patient (κ = 0.47; data not shown), with a 75% concordance.

Detection rate for high grade CIN

High grade CIN was detected in 33 women (46%; table 3). In two of these women the lavage-doctor samples were unsatisfactory for cytological reading. The cytology results from the lavage-doctor would have identified 19 of the detectable 31 high grade CIN lesions (61%), with a specificity, positive, and negative predictive value of 81%, 73%, and 71%, respectively. Seven patients with a high grade CIN lesion did not perform the self sampling. The cytology results in the lavage-patient would have identified 11 of the 26 eligible patients with high grade CIN (42%), with a specificity, positive, and negative predictive value of 93%, 85%, and 65%, respectively, which is not significantly different to the performance of the lavage-doctor (Mc-Nemar χ² = 0.2).

Thirty of 33 (91%) high grade CIN lesions would have been identified when colposcopic examination was performed in the case of a positive high risk HPV test result in the smear. In one woman with a high grade CIN lesion the β globin PCR was negative in the lavage-doctor. Thus, a high risk HPV positive test result in the lavage-doctor would identify 29 women out of 32 (91%) women with high grade CIN lesions, whereas a high risk HPV positive test result in the lavage-patient would identify 21 of 26 (81%) eligible women with high grade CIN. The performance of HPV testing in the smear and in the lavage-patient was not significantly different (Mc-Nemar χ² = 1.3) The specificity for high grade CIN of high risk HPV testing in the smear and in the lavage-patient was 42% and 68%, respectively, with positive predictive values of 58% and 70%, respectively, and negative predictive values of 84% and 79%, respectively.

No significant difference in the detection of high grade CIN could be found between the HPV test results in the lavage-doctor and the lavage-patient (Mc-Nemar χ² = 1.3).

DISCUSSION

Our results indicate that cytological screening for cervical cancer by self sampling is no alternative for cytological screening by the classic Pap smear. The agreement between the Pap smear and the lavage by the patient was low, and less women with high grade CIN would be identified by cytology in the lavage than in the Pap smear. In contrast, high risk HPV testing in self obtained cervicovaginal lavage is a feasible alternative method. The sensitivity for high grade CIN in women with a high risk HPV positive test result in the lavage by the patient was lower, although not significantly lower, than in the smear (81% v 91%). The specificity of the HPV test in self sampled material was higher than in the smear (68% v 42%).

We included women with abnormal and normal cervical cytology to evaluate the use of the self sampling device in these two groups. The self sampling device was acceptable to 88% of the participating women. No differences in acceptability were seen in women with or without abnormal cervical cytology. Seventy seven per cent of the participating women would choose self sampling by vaginal lavage above the classic Pap smear as an alternative screening tool for their next screening round, on condition that both screening methods obtain equal results. Seven women questioned the efficacy of the self sampling. Their main problem was the uncertainty about the amount of fluid they aspirated. However, we only found one β globin negative sample in these samples, indicating that the perception of the efficacy of this method by the women was different to the reality. In future studies we are planning to modify the instructions about using the device.

We included the lavage taken by the doctor as a control for the lavage taken by the patient. No significant difference in the detection of high grade CIN was found between the test results in the lavage-doctor and lavage-patient, indicating that the conditions of the performances—that is, irrigating the cervix directly or indirectly—of the two different lavages did not differ. However, the concordance between HPV testing in the lavage-doctor and the Pap smear appeared to be higher than that in the lavage-patient.

“77% of the participating women would choose self sampling by vaginal lavage above the classic Pap smear as an alternative screening tool for their next screening round, on condition that both screening methods obtain equal results”

Fifteen women (21%) did not perform a home obtained lavage sample. Ten of them had an abnormal cervical smear. In one third of the cases emotional stress with regard to the examination at the colposcopy clinic was the reason for not doing the self sampling at home.

In our study women were asked to obtain a sample at home. This contrasts with other studies that evaluated self sampling under optimal conditions—at the outpatient clinic just after being provided with extensive information.^9,22 In our study a more realistic condition was investigated. Our participation rate of 79% was high when compared with other studies evaluating home obtained self samples. A participation rate of 68% was seen in a study involving 25 women with abnormal cervical cytology who performed a self administered vaginal lavage at home and returned the sample by mail.²³

The high performance of high risk HPV testing in the lavage in our study is in agreement with other studies. Wright et al found a high risk HPV positive test in patient obtained vaginal swabs in 66% of high grade dysplastic cervical lesions or worse in an unscreened population known to have a high incidence of premalignant lesions.²² Nurse obtained swabs revealed an 84% correlation. They concluded that self testing was as sensitive as a Pap smear performed by a health care provider, and proposed self testing for HPV DNA in areas where access to care is limited. In other studies an 85–93% correlation was found for high grade CIN in patient obtained vaginal swabs,^9,24 indicating that self sampling for HPV is also adequate when other techniques are used. In a future study, we are planning to compare the acceptability and efficacy of self sampling by these vaginal swabs with the lavage device.

We found no significant difference in detecting high grade CIN between HPV testing in self sampled lavage material or physician obtained cervical brushes. Moreover, we found a higher specificity and positive predictive value for HPV testing in self sampled material. As long as proper instructions are given to the women, self sampling for HPV DNA testing seems suitable as an alternative screening tool. Although the Pap smear (with additional high risk HPV testing) remains the best screening tool for cervical cancer and its precursors, the high sensitivity for high grade CIN of high risk HPV testing in self sampled material allows us to advise self sampling in women who decline to participate in such programmes because it could largely reduce the risk of cervical cancer associated with not participating in a screening programme.^6,7 In women who do participate the Pap smear remains the best option.