Upper Aerodigestive Tract Squamous Dysplasia: Correlation With p16, p53, pRb, and Ki-67 Expression

Upper aerodigestive tract (UAT) squamous cell carcinoma is a common malignancy worldwide and is associated with significant morbidity and mortality. Upper aerodigestive tract carcinogenesis, manifested histologically as progressive squamous dysplasia, results from cumulative genetic and epigenetic alterations induced by exposure to carcinogenic agents, particularly alcohol and tobacco.1 Oncogenic human papillomaviruses (HPVs) have also been implicated as carcinogenic agents: HPV has been detected in a subset of UAT dysplastic lesions and cancers.1–10 Despite histologic criteria for the diagnosis of dysplasia, applying these criteria is difficult in some cases, fueling the search for diagnostic aids, particularly in identifying molecular correlates of carcinogenesis.

The nature and sequence of the genetic and epigenetic changes in UAT carcinogenesis are gradually being elucidated, specifically in the realm of alterations in cell cycle proteins. Malignant transformation appears to be contingent on disruption of 2 major pathways controlling the cell cycle, namely, pathways featuring tumor suppressors p53 and pRb. A protein activated by DNA damage, p53 induces cell cycle arrest to allow for DNA repair or induces apoptosis if the damage is too severe; pRb regulates the passage of cells from G1 to S phase and is itself regulated by feedback loops involving p16 and cyclin D1 proteins.

Staining patterns for these proteins vary in dysplastic lesions; some demonstrate inactivation, while others show upregulation. Factors contributing to such varying results may include HPV status, histologic subtype, and antibody specificity. Human papillomavirus status in particular appears to be a major determinant of p16 expression pattern in dysplasia; HPV-positive lesions consistently exhibit p16 overexpression compared with HPV-negative lesions.

An additional consideration in analyzing cell cycle protein expression patterns is that, as the phenotypic correlate of a process characterized by cumulative genetic alterations, the expression profile in dysplastic lesions may be correlated with the severity of dysplasia. Indeed, recent studies found an intriguing correlation between height of p16, p53, and Ki-67 immunostaining and degree of dysplasia in oral cavity lesions.11,12 

In order to further elucidate the molecular correlates of UAT squamous carcinogenesis and to investigate the possible role of immunohistochemistry as a diagnostic aid in evaluating dysplastic lesions, the present study examines UAT dysplasia with reference to p16, p53, pRb, and Ki-67 patterns of expression.

Fifty-three formalin-fixed, paraffin-embedded biopsy specimens of oral cavity and pharyngeal squamous dysplasia (23 mild, 12 moderate, and 18 severe/carcinoma in situ) and 13 cases of keratosis/hyperplasia were obtained from the Department of Pathology archives, as identified in a database output file that included cases from 1982 to 1999. Hematoxylin-eosin–stained sections were evaluated for degree of dysplasia according to World Health Organization histologic criteria13 by one of us (S.W.) who was blinded to clinical information. Disagreements with the original diagnosis were resolved by concurrent review by both of us. One lesion was reclassified from the original diagnosis of severe dysplasia to moderate dysplasia.

Five-micrometer sections of formalin-fixed, paraffin-embedded tissue were deparaffinized with xylene and alcohol. Immunohistochemical staining for p16 (E6H4, p16ink4a kit; Dako, Carpinteria, Calif), pRb (Rb1, Dako), p53 (DO7, Dako), and Ki-67 (MIB-1, Dako) was carried out using antigen retrieval by heating the slides in target retrieval solution (Dako) in a steamer set at 95°C for 10 minutes (p16), by heating in a microwave in 10mM Tris-hydrochloride buffer 4 times for 5 minutes each (pRb), or by heating in a microwave in citrate buffer (pH 6.0) 2 times for 5 minutes each (p53 and Ki-67). The slides and buffer were allowed to cool for 20 minutes at room temperature.

For p16, endogenous peroxide was blocked by immersing the sections in the supplied peroxidase blocking reagent. After rinsing in diluted kit buffer, primary antibody diluted 1:25 was applied for 30 minutes at room temperature. After rinsing, the specimen was covered with 200 μL of the kit-supplied visualization reagent for 30 minutes at room temperature. After rinsing, the sections were covered with diaminobenzidine, developed for 10 minutes, and then rinsed.

For pRb, p53, and Ki-67, endogenous peroxide was quenched with hydrogen peroxide solution in water. pRb antibody was used at a 1:25 dilution in Dako diluent for 30 minutes at room temperature. Ki-67 and p53 antibodies were used at a 1:100 dilution in Dako diluent for 15 minutes at room temperature. The slides were rinsed with Dako 1× Wash Buffer twice for 5 minutes. Dako Rabbit Envision HRP System reagent was then applied to the slides for 30 minutes at room temperature. The slides were then rinsed with Dako 1× Wash Buffer twice for 5 minutes. The slides were developed by applying Dako DAB Plus for 3 to 5 minutes until color developed. The slides were counterstained with hematoxylin-eosin. Positive, negative, and normal tonsil controls were performed.

Nuclear staining for pRb, p53, and Ki-67 and nuclear and cytoplasmic staining for p16 were considered positive. Height of positive staining within the epithelium, as measured from the basement membrane, was determined by visual inspection and was coded as 0 for absence of staining or as 1, 2, or 3 according to extension of positivity into the lower, middle, or upper third of the epithelium. The proportion of positive cells within the portion of the epithelium with positive staining was coded as 0 for absence of staining or as 1, 2, or 3 for positivity of up to one third, two thirds, or all of the cells.

Five-micrometer sections of tissue from dysplastic lesions were deparaffinized by 2 additions of xylene with intervening centrifugation, followed by 2 additions of 100% ethanol. The vials were then allowed to dry in a heating block set at 80°C. After drying, tissue was digested with proteinase K. Polymerase chain reactions were performed for B-globin (CTT CTG ACA CAA CTG TGT TCA CTA GC; TCA CAA CCA ACT TCA TCC ACG TTC AC) on all extracted material to assure amplifiable DNA. Polymerase chain reaction for HPV sequences was carried out using a set of degenerate primers for mucosal HPV types (HPV-MY09-[CGT-CCM-ARR-GGA-WAC-TGA-TC]; HPV-MY11-[GCM-CAG-GGW-CAT-AAY-AAT-GG]), which amplifies a 450–base pair product. The polymerase chain reaction amplifications were performed in 50-μL reaction volumes. The reactions contained 5 μL of 10× reaction buffer (Applied Biosystems, Foster City, Calif), 3mM magnesium chloride, 200μM deoxynucleotide triphosphate mixture (Applied Biosystems), 0.5μM labeled forward primer, 0.5μM reverse primer, 2.5 U of Taq polymerase (AmpliTaq Gold, Applied Biosystems), 1 μg of genomic DNA, and distilled water to bring the volume to 50 μL. For the B-globin primers, the cycling parameters were denaturation at 95°C for 10 minutes, then 9 cycles of 60°C for 1 minute, 72°C for 30 seconds, 94°C for 1.5 minutes, followed by 34 cycles of 60°C for 1 minute, 93°C for 1 minute, and then 4°C. For the HPV primers, the cycling parameters were 95°C for 10 minutes, 40 cycles of 94°C for 30 seconds, 55°C for 30 seconds, 72°C for 45 seconds, and then 4°C.

Percentage of cases staining, mean staining height and proportion, and SE were calculated. Statistical analysis was performed using Pearson χ² test and Student t test (STATA release 6.0; StataCorp LP, College Station, Tex). Statistical significance was ascertained at P < .05.

Fifty-three biopsy specimens of oral cavity and pharyngeal squamous dysplasia (23 mild, 12 moderate, and 18 severe/carcinoma in situ) from 43 patients were identified. Patient demographics and biopsy locations are given in the Table. The mean patient age was 65.7 years (age range, 31–87 years); 49% of the patients were male. A prior history of primary invasive head and neck squamous cell carcinoma was reported in 21 (49%) of 43 patients. Eleven (26%) of 43 patients had a prior history of radiotherapy to the head and neck. Concurrent invasive squamous cell carcinoma was associated with 27 (51%) of 53 biopsy specimens; 21 (78%) of the 27 were adjacent to the invasive component.

Thirteen biopsy specimens of oral cavity keratosis/hyperplasia were identified in 13 patients with a mean age of 45.9 years (age range, 30–75 years). None of the patients had a history of prior or concurrent dysplasia or squamous cell carcinoma or had a history of head and neck radiotherapy.

Staining for p16 was nuclear and cytoplasmic in distribution (Figure 1, A), while Ki-67, p53, and pRb staining was nuclear (Figure 2, B through D). Using the grading system already described, pRb staining in normal tonsil epithelium was graded as height and proportion of 2 (positive staining in up to the middle one third of the epithelium and in up to two thirds of cells), while Ki-67 staining in normal tonsil was graded as height and proportion of 1 (positive staining in the lower one third of the epithelium and in up to one third of cells). As seen in Figure 2, A and B, p16 and p53 were not detected in normal tonsil epithelium.

Although staining height of Ki-67 was similar in normal epithelium, keratotic/hyperplastic lesions, and mildly dysplastic lesions, Ki-67 staining height increased with increasing degree of dysplasia (Figure 2, A), being predominantly located within the lower one third of the epithelium in mild dysplasia (mean height, 1.00; 95% confidence interval, 0.82–1.18), within the lower and middle thirds in moderate dysplasia (mean height, 1.50; 95% confidence interval, 1.07–1.92), and within the middle and upper thirds in severe dysplasia/carcinoma in situ (mean height, 2.10; 95% confidence interval, 1.66–2.45). The difference between means was significant between mild and moderate dysplasia (P = .01) and between mild and severe dysplasia/carcinoma in situ (P < .001); however, between moderate and severe dysplasia/carcinoma in situ, the trend did not reach statistical significance (P = .06).

No correlation between degree of dysplasia and percentage of cases staining or proportion of cells staining for Ki-67 was noted (Figure 2, B and C). However, proliferative lesions as a whole (keratosis/hyperplasia and dysplasia) exhibited a higher proportion of cells staining than normal epithelium.

The percentage of cases with p16 staining was greatest (62%) among keratotic/hyperplastic lesions and decreased significantly with increasing degree of dysplasia, from 39% of cases of mild to 5% of cases of severe dysplasia/ carcinoma in situ (Figure 2, C). No significant trend in staining height or staining proportion was observed with p16 in relation to degree of dysplasia, apart from an overall lower height and proportion compared with the other markers tested (Figure 2, A and B).

For pRb and p53, no correlation between degree of dysplasia and percentage of cases staining, staining height, or staining proportion was identified. Keratoses/hyperplasias demonstrated a higher pRb staining height than normal and dysplastic epithelia.

The immunoprofile of dysplastic lesions associated with concurrent squamous cell carcinoma was evaluated. It was not significantly different from the immunoprofile of dysplastic lesions without concurrent carcinoma.

One (2%) of 53 dysplastic lesions was positive for HPV DNA using consensus primers for mucosal HPV types. This case was a biopsy specimen of buccal mucosa that was the only severely dysplastic lesion to exhibit any p16 staining (height, 3; proportion, 3).

This series of dysplastic oral cavity and pharyngeal squamous lesions exhibited increasing height of Ki-67 expression with increasing degree of dysplasia. Ki-67, a nonhistone nuclear protein expressed at all stages of the cell cycle except G0, is a widely used marker of cell proliferation.14 Our results support the model of carcinogenesis that postulates increasing loss of control of cellular proliferation with the accumulation of genetic alterations and correlate well with the findings of previous studies of cell proliferation in dysplastic lesions.11,14–20 Although we quantified Ki-67 positivity by height of staining within the epithelium, methods of quantification in previous studies have included calculating the percentage of positive cells, counting the number of positive cells per unit length of basement membrane, and descriptively categorizing staining as absent, basal, or suprabasal. Regardless of the quantification method used, Ki-67 expression has consistently been shown to increase with increasing degree of dysplasia (and increasing grade of carcinoma). Advantages of grading Ki-67 staining by height are speed and ease, as well as its correspondence to one of the histologic features assessed when grading squamous dysplasia, namely, height of atypia within the epithelium.

Height of p16 staining, by contrast, failed to significantly correlate with degree of dysplasia. However, we did observe a significant reduction in the number of p16-positive cases with increasing degree of dysplasia. Previous immunohistochemical demonstrations of p16 expression in UAT squamous dysplasia have yielded varying results. Some studies have found low expression of p16 in dysplastic lesions,21,22 and these findings have suggested that early inactivation of p16 is a critical event in the process of carcinogenesis. Mechanisms of inactivation of p16 protein include somatic mutation of the encoding gene CDKN2A, homozygous deletion of the gene, or methylation of the gene.1,23 In contrast, other studies have reported overexpression of p16 in dysplasia,6,12,24,25 with 2 studies making particular note of progressively increasing p16 expression with increasing degree of dysplasia.12,24 

Human papillomavirus status appears to be a major determinant of p16 expression pattern in dysplasia; HPV-positive lesions consistently exhibit p16 overexpression compared with HPV-negative lesions. Indeed, the one HPV-positive lesion in this series had strong p16 staining and was the only severely dysplastic lesion to demonstrate p16 staining. The probable mechanism of p16 overexpression in HPV infection is by a normal feedback loop: high-risk HPV 16/18 E7 oncoprotein is known to inactivate pRb26,27; overexpression of pRb regulatory proteins such as p16 may be a compensatory defense mechanism.26 

Two models have thus been advanced to account for observed p16 expression patterns in dysplasia: downregulation due to inactivating mutations and compensatory upregulation due to inactivation of pRb. These models may not be mutually exclusive. In this series of predominantly HPV-negative dysplastic lesions, it was not surprising to find low levels of p16 expression. However, given that p16 is not detected by immunohistochemistry in normal UAT squamous mucosa, our finding that almost 40% of cases of mild dysplasia stained for p16 may be explained by an initial upregulation in p16 expression, serving as a compensatory response to early alterations in cell cycle control occurring during the transition from normal to mild dysplasia. The fact that nondysplastic but proliferative lesions such as keratoses/hyperplasias had the highest rate of p16 positivity offers support for the theory that changes in proliferative potential are an early effect on p16 expression. Our observation that the percentage of cases with p16 staining subsequently decreased with increasing dysplasia may reflect the accumulation of inactivating mutations in CDKN2A. Such inactivation would also account for the failure of p16 staining height to correlate with degree of dysplasia.

Alterations in the p53 tumor suppressor protein are also considered to be a key event in head and neck squamous carcinogenesis. In lesions infected by high-risk HPV types, viral E6 protein induces the degradation of p53 protein.27 In HPV-negative lesions, there is a high incidence of mutations in the p53 gene; these mutations are thought to occur early in carcinogenesis.23,27 Although wild-type p53 protein is rapidly degraded and thus is not detectable in normal squamous mucosa, mutant p53 has a much longer half-life and accumulates in cells, allowing its detection by immunohistochemistry. As expected, we detected no p53 staining in normal oral mucosa but did detect p53 in dysplastic lesions starting as early as mild dysplasia, in keeping with the early timing of p53 mutations. In contrast to previous studies demonstrating an increase in p53 expression with increasing degree of dysplasia,11,15,16,18–20,28 we detected no significant correlation between degree of dysplasia and p53 expression (as measured by staining height), proportion of cases staining, or proportion of cells staining. The presence of p53 in keratotic/hyperplastic lesions without atypia is a possible molecular correlate to the epidemiologic observation that such lesions do have a risk, albeit low, of malignant transformation.29–31 

Disruption of the pRb pathway is another important event in carcinogenesis. High-risk HPV type E7 protein is known to bind to and degrade pRb protein26,27; however, in HPV-negative lesions, alterations to the pRb pathway more significantly involve regulators of pRb such as p16 and cyclin D1, rather than pRb itself.2,23 Our finding of no significant difference in pRb staining patterns among normal, keratotic/hyperplastic, and dysplastic mucosa, nor among different degrees of dysplasia, offers support for this theory.

We detected HPV DNA in 1 of 53 biopsy specimens of dysplastic lesions. Although this 2% positivity rate falls within the wide range of reported infection rates (0%– 80%3–10)in UAT squamous dysplasias, it is at the low end of the range. This may relate to the sites of the lesions represented in this series. In head and neck squamous cell carcinoma, HPV infection is most strongly associated with tumors involving the tonsil.2,32 Extrapolating to premalignant dysplastic lesions, the tonsil is underrepresented in the present study (1;cl53 dysplastic lesions).

In conclusion, in this series of predominantly HPV-negative UAT squamous dysplasias, staining patterns of p53 and p16 in dysplastic lesions, although differing from those in normal mucosa, did not correlate with degree of dysplasia. Moreover, the immunohistochemical profile of keratotic/hyperplastic lesions did not consistently differ from that of dysplastic lesions. Further expression and mutation studies are necessary to characterize the molecular changes associated with the observed immunohistochemical staining patterns. By contrast, height of Ki-67 staining correlated positively with increasing degree of dysplasia. This finding is in concordance with Ki-67 expression patterns as demonstrated by more time-consuming and labor-intensive methods of quantification such as counting the percentage of cells stained or counting the number of positive cells per length of basement membrane. Although the significant overlap between observed staining heights for individual lesions of different degrees of dysplasia may show that Ki-67 height determination lacks sufficiently robust characteristics to stand alone as a test, assessment of staining height, like other immunohistochemical analyses, does show potential as an adjunctive measure for determining degree of dysplasia in upper airway lesions.

This study was supported in part by the Frederic W. Stamler Professorship fund at the University of Iowa (Dr Robinson).