Should LSIL with ASC-H (LSIL-H) in cervical smears be an independent category? A study on SurePath™ specimens with review of literature

Background Cervical smears exhibiting unequivocal features of 'low grade squamous intraepithelial lesion' (LSIL) are occasionally also admixed with some cells suspicious for, but not diagnostic of, 'high grade squamous intraepithelial lesion' (HSIL). Only a few studies, mostly reported as abstracts, have evaluated this concurrence. In this study, we evaluate the current evidence that favors a distinct category for "LSIL, cannot exclude HSIL" (LSIL-H), and suggest a management algorithm based on combinations of current ASCCP guidelines for related interpretations. Methods We studied SurePath™ preparations of cervical specimens from various institutions during one year period. Cytohisto correlation was performed in cases with cervical biopsies submitted to our institution. The status of HPV DNA testing was also noted in some LSIL-H cases with biopsy results. Results Out of 77,979 cases 1,970 interpreted as LSIL (1,523), LSIL-H (146), 'atypical squamous cells, cannot exclude HSIL' (ASC-H) (109), and HSIL (192) were selected. Concurrent biopsy results were available in 40% (Total 792 cases: 557 LSIL, 88 LSIL-H, 38 ASCH, and 109 HSIL). Biopsy results were grouped into A. negative for dysplasia (ND), B. low grade (HPV, CIN1, CIN1 with HPV), and C. high grade (CIN 2 and above). The positive predictive values for various biopsy results in relation to initial cytopathologic interpretation were: a. LSIL: (557 cases): ND 32% (179), low grade- 58% (323), high grade- 10% (55); b. LSIL-H: (88 cases): ND 24% (21), low grade- 43% (38), high grade- 33% (29); c. ASCH: (38 cases): ND 32% (12), low grade- 37% (14), high grade- 31% (12); d. HSIL (109 cases): ND 5% (6), low grade 26% (28), high grade 69% (75). The patterns of cervical biopsy results in cases reported as LSIL-H were compared with that observed in cases with LSIL, ASC-H, and HSIL. 94% (32 of 34) of LSIL-H were positive for high risk (HR) HPV, 1 was negative for HR HPV but positive for low risk (LR), and 1 LSIL-H was negative for HR and LR both. Conclusion LSIL-H overlapped with LSIL and ASC-H, but was distinct from HSIL. A management algorithm comparable to ASC-H and HSIL appears to be appropriate in LSIL-H cases.


Background
Cervical smears may exhibit unequivocal 'low grade squamous intraepithelial lesion' (LSIL) in association with atypical cells cytomorphologically suspicious for, but not sufficient to be interpreted as, 'high grade squamous intraepithelial lesion' (HSIL) [1,2]. This concurrence has been identified recently by many laboratories, but its reporting is not addressed in the 2001 Bethesda System terminology (Bethesda 2001) [3][4][5]. Lack of a standardized method of reporting, however, may affect proper application of American Society for Colposcopy and Cervical Pathology (ASCCP) guidelines based on Bethesda 2001 [6,7]. Although 'atypical squamous cells, cannot exclude HSIL' (ASC-H) is not a definitive interpretation, it is related to an increased risk of higher-grade lesions on biopsy [3]. On the contrary, LSIL is a definitive interpretation with relatively lower prevalence of high-grade lesion on subsequent biopsy.
Currently, the reporting pattern to communicate this concurrence varies amongst different cytopathology laboratories. In our institution, for statistical and quality assurance reasons, we report this concurrence under definitive interpretation as LSIL with a comment that ASC-H cells are also present. Others essentially follow a reverse approach and report it as ASC-H with LSIL in the comment. Some interpreters may choose to combine the associated ASC-H component with LSIL and report the combination as HSIL, which may lead to a potentially high false positivity rate. Rarely, the ASC-H component may be downgraded to 'atypical squamous cells of undetermined significance' (ASCUS) with the final interpretation as LSIL, with potentially false negative results for high-grade lesions.
In the current study, we evaluated cervical biopsies in cases of "LSIL with ASC-H" (LSIL-H) (

Materials and methods
We studied SurePath™ [TriPath Imaging Inc, Burlington, NC] preparations of cervical specimens over a one year period ( Figure 1). Initial cytopathologic interpretations were performed by more than five different cytopathologists, based on Bethesda 2001 criteria and were representative of general cytology reporting patterns [4].
LSIL-H was defined as cases with unequivocal LSIL, in concurrence with ASC-H (Figure 2 and 3). ASC-H cells showed cytomorphologic features reported previously [8]. As observed in a subset of ASC-H, some of these cases showed small atypical parakeratotic (SAPK) cells ( Figure  3), which may demonstrate superficial resemblance to high grade cells in liquid based cytology [9,10].
The cytology results were correlated with concurrent biopsies (both colposcopically-guided and LEEP biopsies) performed at the time of, or within 3 months of cytologic interpretation. Many institutions, in addition to our primary hospital, submitted cervical cytology specimens. Cervical biopsies from some outside institutions were not sent to our institution for surgical pathology and the cytohisto correlation was available in only a fraction of cases. The cervical biopsy results for cases with cytologic interpretation as LSIL, LSIL-H, and ASC-H were compared using the Chi-square test for statistical analysis.
In addition, the status of HPV DNA testing (Hybrid Capture (HC) II, Digene, Silver Spring, MD, USA) performed within a year of biopsy was noted in LSIL-H cases with biopsy results.

Results
We studied 77,979 cases over a one year period ( Figure 1 The biopsy results were grouped into: A. negative for dysplasia (ND), B. low grade (HPV, CIN1, or CIN1 with HPV), and C. high grade (CIN 2 and above) ( Table 1). Positive predictive value (PPV) for each category of biopsy result was calculated for LSIL, LSIL-H, ASC-H, and HSIL interpretations ( Table 1).
LSIL-H (Figure 2 &3) had a lower prevalence of negative biopsy results, compared to ASC-H and LSIL (24% negative results with LSIL-H versus 32% with ASC-H and 32% with LSIL) ( Table 1). LSIL-H had a higher chance of association with high grade dysplasia on biopsy, comparable to that for ASC-H (Positive predictive value [PPV] of 33% with LSIL-H and 31% with ASC-H). LSIL alone was associated with a significantly lower risk (PPV 10%) for high grade dysplasia as compared to LSIL-H (PPV 33%). PPV of LSIL-H (33%) was lower for high grade lesions as compared to HSIL (69%). LSIL-H was associated with higher number of negative biopsy results (24%) as compared to HSIL (5%). However, as compared to ASC-H (32%), the prevalence of negative biopsy results with LSIL-H was relatively lower (24%) (Table 1, Figure 4).
PPVs between different groups were compared by performing two separate 2 × 3 Chi-square tests for LSIL versus LSIL-H and LSIL-H versus ASC-H (Table 1). Applying the Bonferroni correction for multiple testing, the significance level was adjusted to 0.025. The differences between LSIL and LSIL-H were statistically significant (Chi-square value of 35.7 with p value less than 0.001). However, for LSIL-H and ASC-H the difference was not significant (Chisquare is 0.87 with p value nearly equal to 1) (Table 1). Thus, the pattern of biopsy results for LSIL-H overlapped on the lower side with LSIL, but was statistically different. On the higher side, it approached ASC-H, without a statistically significant difference, but slightly higher prevalence of negative results as compared to LSIL-H.
Results of HPV DNA testing (HPVT) performed within one year of the concurrent cervical biopsy were available in 34 of 88 LSIL-H cases. 94% (32 out of 34) of LSIL-H cases were positive for high risk HPV. Two cases were negative for high risk HPV. One was negative for both high and low risk HPV, but the biopsy showed mild dysplasia with HPV. The other case was positive for low risk HPV only, but the biopsy showed high grade dysplasia-(not graded).

Discussion
Lack of a standardized method for reporting LSIL-H generates some questions because of the risk of compromising quality assurance program. More significantly, the proper application of current management guidelines is predominantly based on Bethesda 2001 [6,7] which does not have LSIL-H category. Accumulating evidence, based on our study and a few other studies [1,2,[11][12][13][14] LSIL-H (Figure 2 &3) had an increased risk of high grade dysplasia on biopsy, which was comparable with ASC-H (33% and 31% respectively), but less than that for HSIL (Table 1, Figure 4). This group, however, had a lower chance of a negative biopsy result as compared to the ASC-H group (24% in LSIL-H versus 32% in ASC-H). As compared to LSIL-H, LSIL alone was associated with a significantly lower risk for high grade dysplasia (33% in LSIL-H versus 10% in LSIL). Thus, LSIL-H cytology showed an increased risk of high grade dysplasia on biopsy, with a PPV higher than that for LSIL cytology, but comparable with ASC-H, and distinctly lower than that for HSIL cytology (Table 1, Figure 4). This confirms the intermediate status of LSIL-H, which overlaps on one side with LSIL, and on other side with ASC-H ( Figure 4).
Review of the literature (Table 2) also shows a higher predictive value of LSIL-H for a high grade lesion on biopsy, with overlap on one side with LSIL and on the other with ASC-H, but is distinct from HSIL [1,2,[11][12][13][15][16][17][18]. However, most of these studies are reported as abstracts [11,13,[15][16][17][18] with availability of only partial data for analytical review (Table 2). There are no reports on the use of SurePath liquid based cytology for diagnosis of LSIL-H as reported by this study. Other studies are based on con-  Table 2).

LSIL-H (with CIN2 & HPV in biopsy)
The abstract reported by Booth et al showed association of LSIL-H cytologic interpretation with high grade dysplasia in 45% of cases on biopsy, as compared to 10% in the LSIL group [13]. These researchers evaluated LSIL-H with reference to ThinPrep imaging [16]. Their abstract showed post-imaging increases in the interpretation of LSIL-H, with a higher predictive value as compared to pre-imaging figures (64% versus 23% with statistically significant difference) [16]. Another abstract by Jain et al reported the significance of number of ASC-H cells in association with LSIL, with the conclusion that the PPV for high grade dysplasia on biopsy increases with higher numbers of ASC-H cells in the smear [18]. Additional abstracts reported variable overlap of LSIL-H with ASC-H and LSIL [15,17].
McGrath et al reported 58 cases initially interpreted as mild to moderate dysplasia in conventional cervical smears, with follow-up biopsies showing high grade dysplasia in 59% and low grade dysplasia in 41% [12]. However, this study did not use the current LSIL-H terminology. Based on univariate and multivariate logistic regression analysis, they did not find any specific morphologic features or relationship with the volume of LSIL-  [14] and Underwood et al [15] for conventional smears and ThinPrep.
A subset of LSIL-H is associated with HPV, CIN1 or CIN1 with HPV on biopsy ( Figure 4). As observed in a subset of ASC-H, some of these cases may be unequivocal cases of LSIL with small atypical parakeratotic (SAPK) cells ( Figure  3) which may demonstrate a superficial resemblance to high grade cells in liquid based cytology [9,10,19]. Careful scrutiny with reference to the morphological spectrum of ASC-H reported previously should facilitate improved interpretation even with cytomorphology [8]. Some of the   Cases with concurrent biopsy*

(Performed at the time of cytology or within 3 months of cytology)
(*Cervical cytology specimens were from many institutions in addition to our primary hospital. All institutions did not submit the surgical pathology to us. Because of this cytohisto correlation could not be performed in all the cases.) Comparison of PPV (Table 1, Figure 4).
Chi-square test was applied for statistical comparison.
ASC-H patterns may represent hyperchromatic crowded groups which should be scrutinized carefully for proper interpretation as reported previously [21,22].
The management of LSIL-H cases has not been addressed currently by the ASCCP guidelines [6,[22][23][24][25]. Based on the biopsy pattern in this study and review of the literature, the initial management may be similar to that of LSIL with referral to colposcopy [6,7], but the subsequent approach may be comparable to HSIL (and ASC-H) ( Figure 5). Or, the management may be entirely similar to ASC-H. Some of the issues to be considered while planning management guidelines include: A. How to manage patients with negative or unsatisfactory colposcopy results? B. Would a conservative approach similar to LSIL or ASC-H cases be optimum? C. Should endocervical sampling be obtained if the colposcopic examination is satisfactory? D. Would a cone biopsy similar to HSIL cases ever be indicated in evaluation of LSIL-H category?
As the risk of high grade dysplasia associated with LSIL-H is comparable with ASC-H, (Table 1, Figure 4) and the association with lower incidence of negative results is similar to LSIL, it is reasonable to apply some combination of ASC-H and LSIL management guidelines. Possible application of the HSIL approach at later stages of the algorithm may be needed ( Figure 5). For ASC-H and LSIL, ASCCP recommends initial colposcopy [6,26]. If the colposcopic evaluation in the LSIL-H management algorithm suggested in Figure 4 is negative, the cytology and biopsy material should be reviewed. If the review leads to a change in the interpretation, one should follow the appropriate ASCCP algorithm [6,26]. As compared to the choice of following either the LSIL or ASC-H algorithm, the application of the suggested guidelines will require a relatively small proportion (approximately 1 out of four cases) of already rare LSIL-H cases to go through slightly increased number of clinical encounters ( Figure 5). These guidelines may be evaluated comparatively with other possible alternative combinations to refine it further as indicated by follow up studies in the future.
Another alternative to LSIL-H as a distinct category is to continue with current approach of communicating two distinct interpretations, LSIL and ASC-H, for a given single cervical specimen. Although this approach accommodates 2001 Bethesda System terminology, it has several disadvantages, including difficulties in organizing quality assurance statistics. This, however, may interfere with the management approach based on current ASCCP guidelines [6,[22][23][24][25]. The current recommendation is to refer both LSIL and ASC-H cases to colposcopy, but LSIL-H cases have a significantly lower chance of negative results (Table 1), and so relatively aggressive follow up steps may be indicated at later stages of management. Another challenge is difficulty in verifying the risk of progression to high grade dysplasia for two separate interpretations.
HPV DNA testing has been suggested to be a simple alternative with sensitivity and negative predictive value approaching 100% for detecting HSIL [26][27][28][29]. The role of HPV testing in primary screening of cervical cancer currently has been effective in the ASCUS category. Its role in other categories is evolving. It has also been reported to be helpful in ASC-H cases [27]. However, in LSIL-H, most of the cases are expected to be positive for HPV testing and so its role may be limited. In the current study, coincidental observation of HPV test results were available in 34 out of total 88 LSIL-H cases. HPV testing was positive for high risk HPV in 94% (32 out of 34). Cervical biopsy in two negative cases showed mild dysplasia with HPV in one (negative for both low-risk and high-risk HPV) and high grade dysplasia-not graded in the other (negative for high-risk HPV but positive for low-risk HPV). These findings, although based on small numbers, suggest that HPV testing is not a useful ancillary test in the management of LSIL-H.
Other possible ancillary tests include ProExC [30,31] and p16 INK4A (p16) [19,28,29,[32][33][34][35]. A few have reported role of ProExC with Topoisomerase II alpha (TOP2A) and minichromosome maintenance protein 2 (MCM2) in detection of cervical high-grade squamous intraepithelial lesions from cytologic samples [30,31]. The current literature supports the role of p16 in squamous dysplasia [19,28,29,[32][33][34][35]. The results of p16 on cell block sections of cervical cytology specimens [36] can be interpreted more objectively. The interpretation of the specific nuclear immunoreactivity for p16 is consistent with HPV related dysplasia. Nonspecific cytoplasmic staining may be present in surgical pathology and cell block sections, but this does not interfere with the evaluation of nuclear immunoreactivity. However, in cytology smears, this nonspecific cytoplasmic staining interferes with evaluation of nuclear immunoreactivity with certainty [37,38]. Initial observations suggest that p16 may play a significant role as ancillary test [19,32,37]. The role of these molecular markers is evolving and may help in evaluation of LSIL-H in future.
In conclusion, although LSIL-H as interpretation category is not a unique biologic entity, it correlates with increased risk of high grade dysplasia on biopsy. A pattern of biopsy results, intermediate between LSIL and ASC-H but distinct LSIL-H as category overlap with LSIL and ASC-H, but was distinct from HSIL Figure 4 LSIL-H as category overlap with LSIL and ASC-H, but was distinct from HSIL. As compared to the LSIL interpretations; LSIL-H and ASC-H showed greater association with high grade dysplasia on biopsy. Compared to LSIL and ASC-H; incidence of negative biopsy results was lower with LSIL-H. However, as compared to HSIL group, LSIL-H had higher incidence of negative results. LSIL-H had higher association with low grade lesion than ASC-H.