Assess and manage premalignant lesions of the cervix, including colposcopic assessment

Clinical overview

Cervical premalignancy at this level is about defending a management decision under the twin pressures that define it in South Africa: the largest HIV-driven precancer burden in the world, and a screening programme mid-migration from cytology to HPV-DNA primary testing. The basics are assumed here: the transformation zone and the two stains, HPV carcinogenesis and the LSIL/HSIL terminology, benign mimics of premalignancy, and the SA screening policy and 90–70–90 elimination target. Read those first; this chapter covers the registrar-level decisions — risk-based management, overtreatment, glandular disease, treatment in special populations, and appraising the primary literature behind the algorithm.

The conceptual shift is the move from a test-result paradigm to a risk paradigm. Modern management does not act on a Pap label; it acts on an estimated probability of CIN3+ derived from current and prior results combined. This is the engine of the 2019 ASCCP guidelines, and the central idea in this objective. South African policy applies the same logic with a deliberately pragmatic, single-visit, programme-level overlay because loss-to-follow-up — not test sensitivity — is our rate-limiting failure.

Core knowledge

The two carcinogenic phenotypes, and why subtype changes everything

High-risk HPV drives cervical cancer through E6 (p53 degradation) and E7 (Rb inactivation). The distinction that matters here is that this single mechanism produces two biologically divergent precancers whose detection, treatment and surveillance differ at every step.

• The squamous lineage (CIN/SIL). Arises at the squamocolumnar junction from a discrete population of cuboidal "reserve"/embryonic cells in the original transformation zone. It is visible, contiguous and acetowhite, so colposcopy and ablation work. HPV-16 dominates and is the most transforming genotype — it accounts for the lion's share of squamous cancer and is over-represented in CIN3 relative to its prevalence in low-grade disease, which is why genotype, not cytology grade, best indexes squamous risk. The transformation-zone biology is the reason a Type 3 (endocervical, invisible) TZ forces a different plan: the disease lives where neither stain nor ablation can reach.
• The glandular lineage (AIS → adenocarcinoma). Arises from endocervical columnar epithelium higher in the canal, is HPV-18/45-enriched, and is multifocal with skip lesions. This is not a cosmetic difference: it defeats the three pillars of squamous management. Cytology under-samples it (glandular cells exfoliate poorly and the lesion sits above the os), colposcopy under-reads it (no reliable acetowhite signature, lesions out of view), and ablation cannot treat it (you must have histology and margins). The mechanistic corollary: the rise in cervical adenocarcinoma as a proportion of cervical cancer in HPV-vaccinated and well-screened populations is a screening-failure signature — cytology-based programmes preferentially prevented squamous disease, leaving the glandular fraction behind. HPV-primary screening narrows that gap because 18/45 detection does not depend on exfoliated abnormal cytology.

The genotype anchor is concrete: in ATHENA an HPV-16/18-positive woman carried a 24.4% absolute risk of CIN2+ versus 14.0% for other high-risk types and 0.8% if HPV-negative — which is why 16/18 positivity alone pushes her across the colposcopy threshold regardless of a normal Pap. The corollary, from the ALTS genotype substudy, is the trap: a non-16/18 high-risk-HPV-positive woman is not safe either — her 2-year CIN3+ risk stays ≥7.8%, above the action threshold — so genotype triage escalates 16/18 but does not discharge the rest.

CIN2 is the unstable middle, not a number on a continuum

The clinically load-bearing subtype distinction within squamous disease is that CIN2 is a poorly reproducible, biologically heterogeneous category — a mix of regressing HPV effect and genuine precancer — whereas CIN3 is a reproducible, obligate-treat lesion. This is why p16 immunohistochemistry (the LAST/WHO recommendation) is used to dichotomise an equivocal CIN2 into a low-risk (p16-negative, manage as LSIL) or high-risk (p16-block-positive, manage as HSIL) lesion rather than reflexively excising it. CIN2 in a young woman is the one squamous diagnosis where active surveillance is legitimate because its regression rate is high and cervical-length preservation matters — a judgement that CIN3 never permits.

The risk paradigm and clinical-action thresholds

The 2019 ASCCP Risk-Based Management Consensus Guidelines translate combined results into immediate CIN3+ risk and map that to one of six actions. The excisional thresholds are worth memorising: expedited treatment (treat without a confirmatory colposcopic biopsy) is preferred at an immediate CIN3+ risk of ≥60%, expedited treatment or colposcopy is acceptable at 25–<60%, colposcopy is recommended at 4–<25%, and below that you surveil at 1, 3 or 5 years on the 5-year-risk bands. What matters is not the arithmetic but why this beats a fixed cytology algorithm: it enforces "equal management for equal risk", lets you de-escalate young women with transient infection, and escalates the HPV-16/18-positive woman whose cytology under-reads her disease. The deeper point is what feeds the risk estimate: current result + prior screening history + genotype, so the same HPV-positive/HSIL result carries a different action in a woman with a recent negative HPV history (lower prior, surveil/colposcope) than in one HPV-positive at two consecutive screens (higher prior, toward expedited treatment). A bare Pap label cannot encode that; a risk does.

"Equal management for equal risk" meets a high-prevalence, single-visit system

The WHO 2021 Guideline for screening and treatment of cervical pre-cancer (2nd ed) endorses HPV-DNA as the primary screening test, with a screen-and-treat or screen-triage-treat strategy permitted in the general population, but screen-triage-treat preferred for women living with HIV (WLHIV). Starting ages and intervals diverge by HIV status: general population from age 30, every 5–10 years; WLHIV from age 25, every 3–5 years. South African public-sector policy historically screened HIV-negative women with cytology (three smears, 10-yearly from age 30) and WLHIV at HIV diagnosis and 3-yearly thereafter (2017 NDoH policy); the SASOG/BetterGyn 2024 guideline drove the transition to HPV primary screening with discriminatory genotyping — HPV 16, 18 and 45 carry the highest risk and may be treated without triage, while "other" high-risk types are triaged (cytology/VIA) before treatment. The 2026 NDoH Elimination of Cervical Cancer in South Africa framework now sets the current bands: HPV-DNA primary screening from age 25 for all women regardless of HIV status, repeated every 5 years for life in women living with HIV and every 10 years in HIV-negative women, with a woman under 25 screened only where a particular individual risk exists.

The consultant-level point is why the WHO permits two paradigms and where SA sits between them. Screen-and-treat (treat every screen-positive without triage) maximises programme completion in a single visit — its entire justification is that in a system losing 30–50% of women to follow-up, a confirmed but untreated CIN3 is more dangerous than the overtreatment of a few HPV-positive women who would have regressed. Screen-triage-treat reintroduces a triage step (cytology, VIA, genotype or colposcopy) to cut that overtreatment, at the cost of a second visit and the attrition it invites. SA's discriminatory-genotyping compromise is the elegant middle: it uses the genotype result already in hand from the primary HPV test (16/18/45 → treat; others → triage) as a triage that costs no extra visit, concentrating immediate treatment on the women whose absolute risk justifies it. The decisive trade-off is loss-to-follow-up versus overtreatment, single-visit versus sensitivity.

The glandular problem: AIS is not "HSIL that happens to be glandular"

Adenocarcinoma in situ (AIS) behaves differently and is a recurring trap. It is multifocal with skip lesions, sits higher in the canal, is poorly detected by both cytology and colposcopy, and is HPV-18/45-driven. Excision (not ablation) is mandatory, and margin status is a dominant predictor of residual disease: negative margins leave roughly 6% residual AIS versus ~50–60% with positive/uninterpretable margins. The squamous-disease caveat: in the Arbyn 2017 meta-analysis (97 studies, 44,446 women) involved margins raised residual/recurrent CIN2+ (RR 4.8), but post-treatment high-risk HPV testing predicted treatment failure more accurately than margin status (sensitivity 91% vs 56%) — so a negative test-of-cure HPV, not a clear margin alone, is the strongest reassurance after squamous excision. For AIS the margin retains particular weight (multifocality, skip lesions, poor surveillance access), so a positive AIS margin still mandates re-excision. Fertility-sparing conisation is acceptable in a woman desiring fertility only with negative margins, negative endocervical sampling, and committed long-term colposcopic surveillance; multiple cones are sometimes needed and a positive margin mandates re-excision, not surveillance. Completed childbearing → simple hysterectomy remains standard.

The technical nuance: for AIS the cone should be a single intact, long (tall) cylindrical cone of adequate canal length (at least 10 mm), not a shallow ectocervical loop and not a "top-hat" (separate endocervical) excision, which is unacceptable in AIS because it fragments the specimen. A fragmented specimen makes the margin uninterpretable in a disease where the margin is decisive, and a shallow loop misses canal disease entirely. This is the one premalignant lesion where cold-knife conisation may be preferred over LLETZ — to avoid thermal artefact at the margin in a glandular lesion whose margin must be read.

CIN / SIL grading — recap

Premalignancy carries no FIGO stage — nothing has invaded, so there is nothing anatomical to stage. What you grade instead is the severity of dysplasia: how much of the epithelial thickness is replaced by undifferentiated, abnormal cells. Two systems describe the same squamous disease — the older 3-tier histology grade (CIN 1/2/3) and the 2-tier LAST/WHO terminology (LSIL/HSIL) that maps it onto the biology.

Cytology is not histology — don't conflate the two. The smear/Pap is reported in the Bethesda system along a severity ladder — NILM (negative for intraepithelial lesion or malignancy — a normal result), ASC-US (atypical squamous cells of undetermined significance), LSIL, ASC-H (atypical squamous cells, cannot exclude HSIL), HSIL — with a separate glandular track for AGC (atypical glandular cells; the older term, still used in the SA guideline, is AGUS), AIS and frank carcinoma. That is a screening read of exfoliated cells which triages; the biopsy (CIN 1/2/3, or LSIL/HSIL) is the diagnostic read of tissue architecture that grades. They correspond only loosely — an LSIL smear can sit over a CIN2 biopsy — which is exactly why a high-grade cytology with a normal/low-grade colposcopy is discordance to resolve, not reassure.

Why two histology systems. The 3-tier CIN scheme grades by how far up the epithelium the undifferentiated cells reach (lower third → two-thirds → full thickness). The 2-tier LSIL–HSIL scheme collapses this to match the biology: LSIL = CIN1 = productive viral infection that mostly clears, HSIL = CIN3 (± p16-positive CIN2) = transforming infection that is the true cancer precursor. p16 immunohistochemistry is the tie-breaker for the unstable middle — a block-positive stain reclassifies an equivocal CIN2 as HSIL (treat), a negative stain as LSIL (observe). The glandular lineage (AIS) sits outside this grading entirely: it is not tiered into CIN-equivalents, and its management is driven by margins, not by a grade.