Innovations in Pap Screening for Cervical Neoplasia

, University of California, San Francisco


Medscape General Medicine. 1996;1(1) 

In This Article

Sensitivity of the Pap Smear

Occasionally, the public has been alarmed by media reports of cases of cervical cancer preceded by normal Pap smears (The Wall Street Journal. November 2, 1987:1). To savvy clinicians, this should not be surprising, as the sensitivity of a single Pap smear is reported to be between 60% and 80%.[3] Repeating screens over time increases the Pap's otherwise relatively low sensitivity.

The probability of a false-negative should theoretically equal (0.2)^n, where n is the number of smears performed (if false-negatives are independent). Thus, the probability of 3 consecutive smears being false-negatives should be less than 1%. Using a worst-case scenario--calculating a false-negative rate of 40%--the probability of 3 consecutive false-negative smears is approximately 6%. Yet, the fact remains that some cancers are diagnosed in women who have normal Pap screening. These cancers may be rapidly growing tumors or may represent false-negative Pap smears.

From a different perspective, we can ask what proportion of women with cervical cancer had a recent normal (false-negative) Pap smear. Walker and colleagues[4] reviewed 96 patients with invasive squamous-cell cancer of the cervix. Only 28% of these women had a Pap smear within 5 years of diagnosis of cancer. Janerich and associates[5] examined the screening histories of 481 women with invasive cervical cancer and found that 29% had no prior Pap, 24% had a Pap more than 5 years prior to their cancer diagnosis, and 11% had an abnormal Pap with inappropriate follow-up. Likewise, when Wain and coworkers[6] investigated the Pap histories of 237 patients with cervical cancer in Australia, they discovered that 35% had never had a Pap and 20% were screened more than 4 years before the diagnosis of cancer. Anderson and others[2] reported screening histories of 437 patients with invasive cervical cancer seen from 1985-1988 in British Columbia. Their results revealed that 39% had never had a Pap smear and 10% had not had a smear within 5 years of diagnosis of cancer.

While these studies suggest that a large proportion (49% to 72%) of patients with cervical cancer had not been screened or had been improperly screened, as many as 30% to 50% of women with cervical cancer had had a normal smear within the prior 3-4 years. These findings illustrate not only that there is room for improvement in cervical cancer screening, but also that technical improvements in Pap smears may result in only a limited improvement in mortality from cervical cancer. This does not mean, however, that we should abandon attempts to improve screening techniques.

Some studies have reexamined Pap smears that were read as negative in women who subsequently developed cervical cancer. Walker and colleagues[4] reexamined 15 Pap smears that had been read as normal within 5 years of diagnosis of cancer. Approximately 50% were again read as negative. These results could represent failure of sampling technique, or rapidly growing cancers. Of those read as negative, 25% were believed to be inadequate smears that, if repeated, might have detected the cancer. The remaining 25% had evidence of dysplasia or cancer on the smear and were false-negative because of reading errors.

Rylander[7] reexamined 56 normal Paps in women who developed cervical cancer within 5 years. Thirty-five were reclassified as class II-V, and 14 lacked glandular cells (endocervical cells). In the study of Janerich and associates,[5] only 7% of women had Paps that were misinterpreted as normal. In the study conducted by Anderson and others,[2] only 15% had misinterpretation of previous Pap smear.

Most of these studies suggest that in women who developed cervical cancer, 25% or fewer Pap smears were misread and another 25% may have been inadequate samples. What does all of this imply for screening for cervical cancer?

As a gross estimate, if the results of these studies are applied to the US population, we can project that, of the 16,000 cases of cervical cancer expected to occur, 8000 will have a history of no or inappropriate Pap screening, 4000 will have failure of the sampling technique or rapidly growing cancers, 2000 will have inadequate Pap smears that are not repeated rapidly enough to detect the cancer, and 2000 will have Paps in which abnormal cells are not recognized.

These numbers must be kept in mind when evaluating new methods of cervical cancer screening. About 25% (or 2000 per year) of the 8000 false-negative Paps are potentially preventable by refinement of Pap reading, and an additional 25% by improvement in the sampling technique/process. On the other hand, an equal number of cases of cancer will not be detected by the Pap smear.