The probability for a Pap test to be abnormal is directly proportional to HPV viral load: results from a Swiss study comparing HPV testing and liquid-based cytology to detect cervical cancer precursors in 13,842 women

Abstract

In a study involving 13,842 women and 113 gynaecologists, liquid-based cytology and HPV testing for detecting cervical cancer were compared. A total of 1334 women were found to be positive for one or both tests and were invited for colposcopy with biopsy. A total of 1031 satisfactory biopsies on 1031 women were thereafter collected using a systematic biopsy protocol, which was random in the colposcopically normal-appearing cervix or directed in the abnormal one. In all, 502 women with negative tests were also biopsied. A total of 82 histologic high-grade squamous intraepithelial lesion (HSIL) were reported in biopsies, all from the group with one or both tests positive. Sensitivity and specificity to detect histologic HSIL were 59 and 97% for cytology, and 97 and 92% for HPV. In total, 14% of reviewed negative cytological preparations associated with histologic HSIL contained no morphologically abnormal cells despite a positive HPV test. This suggested a theoretical limit for cytology sensitivity. HPV viral load analysis of the 1143 HPV-positive samples showed a direct relationship between abnormal Pap test frequency and HPV viral load. Thus, not only does the HPV testing have a greater sensitivity than cytology but the probability of the latter being positive can also be defined as a function of the associated HPV viral load.

Figure 1

Percentage of HPV infection (y-axis) against age intervals (x-axis). 95% confidence intervals are superposed to grey bars. The highest HPV infection rates are found in the youngest women. The percentage progressively declines as age increases (n=13 840).

Figure 2

Percentage (y-axis) against HPV viral load of 1143 HPV+ samples (x-axis) reported as Logarithm value of Relative Light Unit (LnRLU). The ‘% women HPVPOS’ curve shows the distribution of the HPV+ samples among LnRLU intervals. In this study, most HPV+ samples have low LnRLU: near 50% of HPV+ samples are found in the first three LnRLU intervals. The grey bars represent the percentage per LnRLU interval of positive cytological preparation (⩾ASC-US); it shows a direct relationship between the HPV viral load and the probability for a Pap test to be positive. The HSIL CYTONEG and LSIL CYTONEG curves represent the percentage per LnRLU interval of false negative cytological preparations with subsequent biopsy proven lesion.

Figure 3

Retrospective analysis of all cytological preparations associated with subsequent histological HSIL. Logarithm values of RLU (y-axis) against number of positive cellular spots (x-axis) retrospectively counted on cytological preparations. False negative CYTO (black triangles) have a smaller viral load and fewer number of positive cellular spots than true positive CYTO (white squares). Both differences are statistically significant. However, a significant number of false negative cytological preparations (11 out of 34 (32%)) have no cellular spots despite a significant viral load (n=79).