Are Fluid-Based Cytologies Superior to the Conventional Papanicolaou Test? A Systematic Review

Study Quality Assessment

The total quality assessment scores of the 5 studies ranged from 7 to 10 out of a possible maximum score of 13. In general, the studies offered little or no description of the clinical characteristics of the women screened. Most did not even mention the women’s ages. All of the studies included at least some women who were at high risk for an abnormal Pap test result. Two of the studies included only women referred because of a previously abnormal result. Three studies included colposcopy clinics among their recruiting sites. One study was done in Costa Rica because of the known high prevalence of abnormal Pap test results in that country. One study specified that a consecutive series of patients was used, and one used a population-based random sample. The other 3 studies gave no details of how patients were selected for inclusion in the trial. All studies used prospectively collected data, and the methods used to actually perform the Pap and FBC tests were universally well described. Two of the studies used methods to minimize the effects of interobserver reliability in the evaluation of the test results, but none addressed this issue as it related to the assessment of the reference standard.

Sensitivity and Specificity

The ROC curves for FBC and Pap using all 5 of the above studies are displayed in Figure 1. The AuROC curves were similar (Pap=0.93; FBC= 0.91). This difference was not significant (P=.37), and the confidence interval (CI) was wide (95% CI, -0.33 to 0.80). FBC demonstrated higher sensitivity, 90% (95% CI, 0.77-0.96) versus 79% (95% CI, 0.59-0.91) for Pap. FBC had a lower specificity, 85% (95% CI, 0.74-0.92) versus 89% (95% CI, 0.75-0.96) for Pap.

Specimen Adequacy

FBC specimens were more likely to be reported as satisfactory (RD=0.06; 95% CI, 0.03-0.09; Figure 2. There was no significant difference in the number of unsatisfactory test results. There was a 6% higher rate of absence of endocervical cells (RD=0.06; 95% CI, 0.02-0.10) but a 10% decrease in reports of SBLB-o (RD = -0.10; 95% CI, -0.13 to -0.06) with FBC. The increase in absence of endocervical cells for FBC specimens was seen in all split-sample studies (RD= 0.08; 95% CI, 0.06-0.11) but not in the cohort studies (RD = -0.01; 95% CI -0.07 to 0.05).

Discussion

The use of FBC increases both true-positive and false-positive results when compared with Pap. These conclusions must be considered tentative because of the lack of statistical significance and the significant methodologic problems found in the studies used. Two recent meta-analyses have addressed the accuracy of FBC tests.^3,60 Neither addressed the issue of specimen adequacy. Although neither included all 5 studies that we identified for our analysis of test accuracy, our findings of a possible increase in sensitivity with a decrease in specificity are consistent with the findings of the other 2 meta-analyses.

The possibility that FBC may increase sensitivity but decrease specificity should lead to caution in the adoption of this technology, especially for women at low risk for cervical cancer. For women with no history of abnormal findings on previous Pap tests (estimated prevalence of cervical cancer = 0.05%) more than 1800 FBC would be needed to detect one additional true positive. More than 50 additional false-positive test results (“cancer scares”) would accompany each additional true positive. The costs of follow-up investigations for these additional false-positive tests must be added to the additional cost of the test itself in assessing the potential impact of a widespread switch to this new technology. Many women are at risk because they fail to obtain regular Pap tests, often because of lack of insurance and cost barriers. Unfortunately, the higher cost of FBC may make these women even less likely to have screening performed. Thus widespread use of FBC could, paradoxically, lead to an increase rather than a decrease in cervical cancer deaths by decreasing the use of this important test by lower income women.

The trade-off between sensitivity and specificity is more favorable for women with higher disease rates. In a population with a 3% prevalence of cervical disease, only 300 FBC Pap tests would be required to detect one additional abnormality, and only 16 additional false-positive tests would result. Thus, for women who have had prior abnormal Pap test results or are known to be infrequent attenders, there may be a role for FBC.

Pap test results reported as less than satisfactory can present a significant problem with increased office return visits, increased psychologic trauma to the patient, and increased costs of repeated tests. Thus the decrease in “SBLB other” reports is a benefit of FBC, although it is partially offset by an increased absence of endocervical cells. Some have suggested that the absence of endocervical cells is an artifact of the study methods for split-sample studies (where the collection instrument is first wiped across a slide for the Pap and then inserted into the FBC vial). We do not understand why this process would preferentially extract endocervical cells. We did note that the cohort studies showed no increase in SBLB absence for FBC. However, in these studies, the Pap and FBC specimens were collected from different women and often involved different time periods and different physicians using different collection instruments. Both the degree of heterogeneity and the reported differences on all comparisons between Pap and FBC Pap tests were consistently larger in the cohort than in the split-sample studies.