Integrative analysis of gene alterations and immunoexpression profiles of cell cycle checkpoints in oral squamous cell carcinoma

Abstract

BACKGROUND:

Oral squamous cell carcinoma (OSCC) represents 95% of all cancers of the head and neck region. The five-year survival rate of OSCC patients is about 60% and has not gone throw significant improvements despite recent advances in molecular biology, or the identification of key pathways in its pathophysiology such as cell cycle.

OBJECTIVE:

1) to analyse the inmunoexpression of cell cycle checkpoints (CPs) in an OSCC institutional cohort and to relate it to clinicopathological features and survival, and 2) to study CPs-related genes in the OSCC subset of the TCGA database.

METHODS:

Immunohistochemistry (IHC) for p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ protein expression was quantified by tissue microarray analysis in 68 samples from OSCC patients. In order to analyse its correlation with genetic information, a cohort belonging to The Cancer Genome Atlas (TCGA) database was analysed.

RESULTS:

Of 68 patients, 34 (50%) developed recurrence, and 36 (52.09%) died as a result of disease progression (mean survival 34.09 months). IHC staining for nuclear cyclin D1 was associated with worse staging and tobacco use. p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1, and p27 ${}^{\text{KIP1}}$ expression was unrelated to overall survival. No statistically significant correlation linked the CPs-related genes mutations to OSCC overall survival in the TCGA database.

CONCLUSIONS:

CPs variations at a phenotype and genotype level seem not to affect significantly clinicopathological features and survival in the studied OSCC cohorts.

1. Introduction

Carcinogenesis is an intricate multistep process initiated by the accumulation of genetic variations in proto-oncogenes and tumour suppressor genes (TSGs) [1]. Cells have evolved strategies to prevent those hazardous genetic events. Nonetheless, those machineries are not always able to repair genomic instability triggering cell-cycle defects, angiogenic switch or senescence which ultimately results in a cancer cell generation and preservation [2]. The cell cycle or division is the process in which cell duplicates its contents and divides, resulting in two genetically identical daughter cells. Cell division is monitored by cell cycle checkpoints (CPs). The canonical CPs are cyclins and cyclin-dependent kinases (CDK), these complexes are inhibited by CDK inhibitors (CKIs). These proteins are classified as TSGs due to their ability to disrupt cell proliferation, and according to a bulk of molecular oncology research their coding genes are located in altered chromosomal areas across multiple human cancers [3]. Based on structural and functional characteristics, there are two CKIs families: the CIP/KIP family (including p21 ${}^{\text{Cip1}}$ , p27 ${}^{\text{KIP1}}$ , and p57), and the INK4 family (including p15, p16 ${}^{\text{INK4a}}$ , p18, and p19) [4].

CIP/KIP proteins have a broader spectrum of substrates, binding to cyclin D, cyclin E and CDKA-1 complexes, so having greater influence in carcinogenesis. P21 ${}^{\text{CIP1}}$ is activated under stress conditions and can follow two pathways: one dependent on p53, and another independent of p53 stimulated by cell growth factors [5]. The p27 ${}^{\text{Kip1}}$ was the first CDK inhibitor discovered and its related gene is located on chromosome 12p-12-12p13.1. Both aforementioned proteins are responsible for blocking the activity of the cyclin E/CDK2 and cyclin D/CDK2 complexes, inhibiting phosphorylation of pRb and preventing the cellular transition from phase G1 to phase S, which triggers apoptosis. P16 ${}^{\text{INK4a}}$ is another relevant TSG belonging to the INK4 family whose coding gene locus is located at MTS-1 on chromosome 9p. During the G1 phase, p16 ${}^{\text{INK4a}}$ controls cell cycle by inhibiting the ability of the cyclin D/CDK4/6 complexes to phosphorylate pRb. The phosphorylation of pRb leads to the dissociation of the pRB/E2F complex, which leads to cell cycle progression. On the other hand, the release of E2F activates the transcription of p16 ${}^{\text{INK4a}}$ . The increase of p16 ${}^{\text{INK4a}}$ levels, results in CDK4/6 blocking, which ultimately triggers cyclin D/CDK4/6 complex inhibition. CPs variations have been classically related to human tumorigenesis due to its amplification or overexpression can lead to some hallmarks of cancer [5, 6, 7].

Oral squamous cell carcinoma (OSCC) (ICD 01-10) 1 is the 8 ${}^{\mathrm{th}}$ most common cancer among men and the 14 ${}^{\mathrm{th}}$ most common among women in the US. The overall five-year survival rate of OSCC affected patients has been for decades about 50%, despite advances in therapeutic and aids to prevent diagnostic delay. This standing poor prognosis is associated to the existence of suboptimal anticancer strategies characterised by severe treatment-associated sequels and acquired multidrug resistance leading to recurrences and second primary tumours [8].

Our group has previously unravelled the implications of several CPs in OSCC by means of comprehensive reviews [9, 10, 11], and immunohistochemistry reports [12, 13]. Nonetheless, in current literature there is a lack of studies evaluating multiple CPs altogether constructing a panel in a single cohort of OSCC-affected individuals. Moreover, literature lacks bioinformatics analysis able to integrate available cross-platform data on the implications of CPs in OSCC despite the existence of genomic and molecular profiling datasets like The Cancer Genome Atlas (TCGA) [14]. Prompted by discussed literature, the research hypothesis is that this CPs panel may provide a better characterization of tumour biology and a stronger prognostic value on survival than the use of single markers.

Figure 1.

Immunohistochemistry staining examples in cyclin D1 negatively (A) and positively (B) stained, p16 ${}^{\text{INK4a}}$ negatively (C) and positively (D) stained, p21 ${}^{\text{CIP1}}$ negatively (E) and positively (F) stained, p27 ${}^{\text{KIP1}}$ negatively (G) and positively (H).

The objective of this study was twofold: 1) to assess the predictive impact of a immunohistochemical markers panels, specifically of CPs, in an OSCC well defined cohort and with minimum five-year follow-up relating it to clinical and pathological features and 2) to integrate this information with TCGA genomic database.

Table 1

Association between p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ staining expression with the clinicopathological parameters

Clinicopathological	Variables		p16INK4a			p21Cip1			p27Kip1			Cyclin D1
information		Total $n$ (%)	$n$	Positive (%)	$p$ value	$n$	Positive (%)	$p$ value	$n$	Positive (%)	$p$ value	$n$	Positive (%)	$p$ value
Age
	$<$ 55 years	14 (25.59)	13	5 (38.46)	0.728	12	5 (41.67)	0.356	9	8 (88.89)	0.310	9	3 (33.34)	0.803
	$>$ 55 years	54 (79.41)	55	17 (30.91)		50	14 (28%)		29	21 (72.41)		29	11 (37.93)
Gender
	Male	33 (48.53)	31	13 (41.94)	0.217	30	9 (30)	0.915	15	12 (80)	0.666	24	6 (25)	0.557
	Female	35 (51.47)	33	9 (27.27)		32	10 (31.25)		23	17 (52.17)		14	8 (57.14)
Tumour location
	Tongue	6 (8.82)	5	1 (20)	0.803	5	4 (0.8)	0.204	4	3 (0.75)	0.098	3	2 (1.5)	0.652
	Gingiva	17 (25)	17	7 (41.18)		16	4 (25)		6	5 (83.33)		7	4 (57.14)
	Retromolar trigone	6 (8.82)	6	2 (33.34)		6	1 (16.67)		4	4 (100)		4	1 (25)
	Lip	22 (32.35)	22	8 (36.36)		19	5 (26.32)		14	10 (71.43)		14	4 (28.57)
	Floor of the mouth	8 (11.76)	7	1 (14.29)		7	3 (42.86)		4	1 (25%)		4	1 (25)
	Palate	8 (11.76)	7	3 (42.86)		8	2 (25)		6	6 (100)		6	2 (33.34)
Differentiation
	Well	30 (44.12)	29	9 (31.03)	0.090	27	9 (52.94)	0.922	15	11 (73.33)	0.408	15	6 (40)	0.913
	Moderate	30 (44.12)	28	8 (28.57)		28	8 (28.57)		18	13 (72.22)		18	6 (33.33)
	Poor	8 (11.76)	7	5 (71.43)		7	2 (28.57)		5	5 (100)		5	2 (40)
T stage
	I	20 (29.41)	18	8 (44.44)	0.473	17	4 (23.53)	0.884	12	11 (91.67)	0.176	12	3 (0.25)	0.095
	II	12 (17.65)	12	4 (33.33)		12	4 (33.33)		8	4 (50)		8	1 (0.125)
	III	9 (13.24)	8	1 (12.5)		7	2 (28.57)		1	1 (100)		1	1 (100)
	IV	27 (39.71)	26	9 (34.62)		26	9 (34.62)		17	13 (76.47)		17	9 (52.94)
N stage
	0	55 (80.88)	53	19 (35.85)	0.774	50	18 (36)	0.144	36	27 (75)	0.418	36	13 (36.11)	0.692
	1	7 (10.29)	6	2 (33.34)		6	1 (16.67)		2	2 (100)		2	1 (50)
	2b	6 (8.82)	5	1 (20)		6	0		0	0 (0)		0	0 (0)
Stage
	I $+$ II	32 (47.06)	30	12 (43.33)	0.373	29	8 (42.11)	0.624	20	15 (75)	0.841	20	4 (20)	0.023*
	III $+$ IV	36 (52.94)	34	10 (29.41)		33	11 (33.33)		18	14 (77.78)		18	10 (55.56)
Tobacco use
	Non-smoker (never)	32 (47.06)	30	13 (43.33)	0.286	27	8 (29.63)	0.902	15	12 (80)	0.907	14	6 (42.86)	0.027*
	Former smoker ( $>$ 10 years)	12 (17.65)	11	2 (18.18)		11	4 (36.36)		8	6 (75)		9	6 (66.67)
	Current smoker ( $⩾$ 5 years)	24 (35.29)	23	7 (30.43)		24	7 (29.17)		15	11 (73.33)		15	2 (13.33)
Alcohol use
	Non drinker	33 (48.53)	30	13 (43.33)	0.347	28	8 (28.57)	0.620	16	13 (81.25)	0.621	15	7 (46.67)	0.515
	Non-current drinker	22 (32.35)	21	6 (28.57)		21	8 (38.09)		17	13 (76.47)		18	6 (33.33)
	Current drinker	13 (19.12)	13	3 (23.08)		13	3 (23.08)		5	3 (0.6)		5	1 (20)
Recurrence
	No	34 (50)	31	14 (45.16)	0.078	30	7 (23.33)	0.227	18	15 (83.33)	0.454	18	5 (27.78)	0.272
	Yes	34 (50)	33	8 (24.24)		32	12 (37.5)		20	14 (70)		20	9 (45)
Adjacent dysplasia
	No	38 (55.88)	35	13 (37.14)	0.255	34	12 (35.29)	0.662	20	15 (75)	0.150	20	7 (35)	0.297
	Mild	8 (11.76)	8	1 (12.5)		7	3 (42.86)		4	2 (50)		4	3 (75)
	Moderate	4 (5.88)	4	3 (75)		4	1 (25)		3	3 (100)		3	0 (0)
	Severe	1 (1.47)	1	0 (0)		1	0 (0)		1	0 (0)		1	0 (0)
	In situ	17 (25)	16	5		16	3 (18.75)		10	9 (90)		10	4 (0.4)

*Stands for statistically significant.

2. Material and methods

2.1 Patients

The present retrospective cross-sectional study involved a cohort of 68 patients diagnosed with OSCC surgically treated at the University Hospital of Santiago de Compostela, Galicia, Spain, between April 2005 and March 2010. The study was performed in accordance with the Declaration of Helsinki of the World Medical Association, approved by the Ethics Committee of the Servizo Galego de Saúde (ID: 2013/541) as well as following the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines [15]. For patients included in this study, the staging was stablished according to the 7th American Joint Committee on Cancer (AJCC). The mean follow-up period was 48.55 $\pm$ 47.29 months with a minimum of 2 months and a maximum of 211.81 months. Some of clinical and demographic data were published in our previous report [12].

2.2 Tissue microarray (TMA) generation

Paraffin blocks were selected on the basis of the availability of suitable formalin-fixed, paraffin-em- bedded tissue (at least 1-mm thick). For characterization of immunohistochemical protein expression, we constructed three different TMAs containing representative areas of every tumour. After microscopic evaluation, two areas of each tumour were selected, avoiding necrosis and keratin pools. Each tissue array was assembled as previously described. The areas chosen for the cylinder core were representative of the tumours. For each inmunolabelling run, a positive (tonsil) and negative (omission of primary antibody) controls were used to ensure quality of staining.

2.3 Immunohistochemistry of p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$

Three-micrometer tissue sections from the TMA blocks were sectioned and applied to special immuno- histochemistry-coated slides (DAKO, Glostrup, Denmark). For the immunohistochemical study the following antigen retrieval, and clones were used according to the instructions of manufacturer: p16 ${}^{\text{INK4a}}$ (Dako, FLEX, RTU; clone E6H4), p21 ${}^{\text{CIP1}}$ (Dako, FLEX, RTU, Clone SX118), cyclin D1 (Dako, FLEX, RTU clone EP12), and p27 ${}^{\text{KIP1}}$ (Dako, FLEX, RTU, Clone SX53G8). Epitope retrieval was performed in 10 mM sodium citrate buffer (pH 6.0) using a water bath for 40 min at 95–99 ${}^{\circ }$ C. Endogenous peroxidase was blocked using peroxidase-blocking reagent (DAKO) for 5 min. Incubation with the primary antibody was performed at room temperature for 20 min, and staining was revealed with EnVision DAB (Dako). Isotype controls were used as negative controls in place of primary antibodies.

2.4 Evaluation of immunohistochemical results

Two of the authors (JSP and CCP) performed an observational quantification analysis in a joint session. For quantification, slides were digitized using an automate slides scanner to produce high-resolution images for visual quantitative analysis using an automatic image analysis system ACISRIII (Dako, Glostrup, Denmark). The digital images of the slides were captured by ACIS scanner at low power magnification and the entire slide images were viewed on a monitor. The region of highest immune percentage of positive cells were selected for further automatic scoring. A minimum of three of these areas containing only tumoral cells was selected manually for quantitative evaluation. The final score was the average result of the different areas measured. To confirm the accurateness of the measurements, selected areas from ten of the specimens were measured three times.

Staining evaluation was restricted to the nuclei for p21Cip1 and cyclin D1. Nonetheless, any degree of cytoplasmic or nuclear staining was considered for p27Kip1 and p16INK4a evaluation (Fig. 1). The expression of markers was semiquantitatively evaluated on the basis of the following thresholds as cut-off points for positivity or not: 5% (p21 ${}^{\text{CIP1}}$ and p16 ${}^{\text{INK4a}}$ ), and 50% (p27 ${}^{\text{KIP1}}$ , and cyclin D1). These cut-offs were determined for each marker based on the mean of its expression score.

2.5 Bioinformatics analyses

All the raw data available from TCGA was processed by a public bioinformatics tool (www.cbiopor tal.org). By this platform, only OSCC-affected patients were extracted from the Head and Neck Cancer TCGA database (i.e., 172 patients/samples) in order to study genetic variations related to the previous evaluated CPs. Retrieved information was used to perform statistical analyses of correlation with survival [14].

2.6 Statistical analysis

Statistical analyses were performed by using SPSS statistical software 22.0 for Windows. In order to evaluate the correlation of marker expression with clinic-pathological variables, Spearman correlation test was used. Gaussian distribution of variables was checked through Kolmogorov-Smirnov normality test. The non-parametric test of ${\chi }^2$ was used in order to investigate the difference in expression between groups. Univariate Kaplan-Meier method was used to estimate the overall survival rates by means of the log-rank test. All the reported p values were 2-sided, and considered significant under 5%.

Table 2

Association of p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ staining in the present oral squamous cell carcinoma (OSCC) cohort

	$n$	p16 ${}^{\text{INK4a}}$ positive (%)	$p$ value	$n$	p21 ${}^{\text{CIP1}}$ positive (%)	$p$ value
Cyclin D1			0.361			0.010*
Negative	23	8 (34.78)		24	4 (16.67)
Positive	14	7 (50)		14	8 (57.14)
p27 ${}^{\text{Kip1}}$			0.039*			0.289
Negative	9	1 (11.11)		9	4 (44.44)
Positive	28	14 (50)		29	8 (27.59)

*Stand for statistically significant.

Figure 2.

Kaplan-Meier curves of OS for the staining of p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ in the present OSCC cohort.

3. Results

3.1 Analysis of p16 ${}^{\text{INK4a}}$ , p21 ${}^{\text{CIP1}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ expression in TMAs of OSCC mucosa samples

Clinicopathological characteristics and its association with semiquantitative protein expression is displayed in Table 1. Briefly, the majority of the patients were older than 55 and the distribution regarding gender was almost equal (48.53 vs. 51.47). In terms of classification, tumours have been grouped according to initial stage tumours (i.e., I–II) with 32 cases (47.8%) and advanced stage tumours (i.e., III–IV), with 36 cases (52.9%), most of the tumours did not show an adjacent dysplasia (55.88), nor regional lymph nodes metastasis (80.88%). None of the tumours presented distant metastasis. In stages I–II, 4 (20%) of tumours were positive for Cyclin D1, whereas in advanced stages 10 were (i.e., 55.56%) ( $p=$ 0.023). The overexpression of this protein was also a fact among former and current smokers vs. never smokers as checked by the bonferroni correction ( $p=$ 0.023).

Table 3

Mean overall survival (OS) time by means of Kaplan–Meier method in each selected prognostic factors

Variable	$n$	Univariate analysis overall survival months $\pm$ SD	$p$ value*	HR
Age
$<$ 55 years		35.942 $\pm$ 10.299	0.751	0.101
$>$ 55 years		33.630 $\pm$ 5.736
Gender
Male		26.016 $\pm$ 5.845	0.104	2.651
Female		41.880 $\pm$ 7.845
Tumour stage
I $+$ II		48.971 $\pm$ 8.905	0.006*	7.445
III $+$ IV		22.573 $\pm$ 4.726
p16 ${}^{\text{INK4a}}$
Negative		34.021 $\pm$ 6.418	0.593	0.286
Positive		26.392 $\pm$ 6.744
p21 ${}^{\text{CIP1}}$
Negative		34.738 $\pm$ 6.780	0.646	0.211
Positive		28.681 $\pm$ 6.778
p27 ${}^{\text{KIP1}}$
Negative		29.189 $\pm$ 7.035	0.632	0.229
Positive		27.798 $\pm$ 4.392
Cyclin D1
Negative		32.633 $\pm$ 4.452	0.233	1.424
Positive		26.431 $\pm$ 6.525

* $p$ Log-rank test; HR, hazard ratio SD, standard deviation; CI, confidence interval. *Stands for statistically significant.

Figure 3.

(A) Profiled mutation for CDKN2A, CDKN1A, CCND1 and CDKN1B in each OSCC-affected patient included in TCGA. (B) Network involved in the expression of the CPs genes. Dark circles represent selected genes and an interaction legend is available on the right corner.

The most frequent coexpressions detected in the present cohort are represented in Table 2. When Spearman’s test between markers was applied two strong correlations were detected, once between p27 ${}^{\text{KIP1}}$ -p16 ${}^{\text{INK4a}}$ ( $R=$ 0.340; $p=$ 0.04) and the other between p21 ${}^{\text{CIP1}}$ -Cyclin D1 ( $R=$ 0.420; $p=$ 0.009).

Mean overall survival (OS) time was 34.09 $\pm$ 37.01 months. Tumours in advanced stages showed a lower survival (22.57 months; 95% CI 13.31–31.84) than those in early stages (48.97 months; 95% CI 31.52–66.42) (HR $=$ 7.45, $p=$ 0.006). Intensity of staining for any IHC marker was not significantly associated with lower OS according to Kaplan-Meier analysis (Fig. 2).

3.2 Analysis of CDKN2A, CDKN1A, CCND1 and CDKN1B mutations in OSCC from The Cancer Genome Atlas (TCGA) database

Mutations in aforementioned genes from the 172 TCGA cohort are shown in Fig. 3. The most frequent mutated genes were CDKN2A affecting 90 patients (52%) and CCND1 affecting 26 patients (15.12%), specifically the most frequent variations in CDKN2A were truncating mutations and deep deletions, whereas in the case of CCDN1 gene mutations were mostly amplifications (Fig. 3A). Network involved in the expression of selected CPs is displayed in Fig. 3B.

In an OS analysis within TCGA database, OSCC cases with alterations in selected genes reached a total of 65 cases of which 31 died with a 17.91 median month survival. In the other hand, OSCC cases without alterations in aforementioned genes reached a total of 37 cases of which 16 deceased with a 28.29 median month survival. This OS tendency was not statistically significant by means of Kaplan-Meier method (log-rank test $p$ -value $=$ 0.414).

4. Discussion

The oncogenic role of aberrant signalling pathways that regulate cell division has been investigated in various malignancies including OSCC [1, 2, 3, 4]. The aim of this study was to identify the inmunoexpression signature of four canonical CPs in our institutional database and to integrate these data with genetic information available from TCGA project. Available data gathered from literature presents certain gaps of knowledge regarding the implications of these markers expression towards cancer-related features and survival. To the best our knowledge, only three meta-analyses has been previously published by pooling data from studies evaluating the significance of three out of the four selected CPs (p16 ${}^{\text{INK4a}}$ , cyclin D1 and p27 ${}^{\text{KIP1}}$ ) as a prognostic factor in OSCC patients [16, 17, 18]. Sedghizadeh et al. provided a pooled analysis regarding OS hazard ratios for p16 ${}^{\text{INK4a}}+$ patients versus p16 ${}^{\text{INK4a}}$ -concluding a favourable survival rate for p16 ${}^{\text{INK4a}}+$ (HR $=$ 0.360; 95% CI 0.258–0.503; $p<$ 0.001) [16]. A strong correlation between human papilloma virus (HPV) infection and p16 ${}^{\text{INK4a}}$ protein expression has consistently been reported in head and neck cancers. This molecular event is triggered by pRb inactivation by the HPV E7 oncoprotein. In this vein, p16 ${}^{\text{INK4a}}$ protein expression is part of the current medical gold standard algorithm to stablish the presence of an HPV-induced carcinogenesis in head and neck and tumours [12]. Curiously, Ang et al. demonstrated that p16 ${}^{\text{INK4a}}$ expression is a stronger independent prognostic factor for OS than the presence of HPV DNA [13]. CDKN2A is the p16 ${}^{\text{INK4a}}$ related gene, is located on chromosome 9p21, and has three exons. The most frequent mutations detected on this gene according to TCGA are deep deletions [15] (Fig. 3). Padhi et al. verified that this loss of function has been linked per se to a poor prognosis and survival rate in OSCC [19]. The present report failed to ascertained the implications of this protein as a valuable tool for OSCC prognosis. Scully and Bedi reported that ethnic and geographical difference in oral cancer affected populations could induce an accentuated risk of bias for result interpretation [20]. Taking in account this hypothesis two close studies emerge. González-Moles et al. observed that 41.67% of the tumours studied had negative expression for both of p21 ${}^{\text{CIP1}}$ and p16 ${}^{\text{INK4}}$ proteins but do not found significant difference when they associated their expression with clinical-pathological parameters, including recurrence and survival [21]. On the other hand, Monteiro et al. do found p16 ${}^{\text{INK4}}$ as a useful tool for prognosis when studying cancer-specific survival ( $P=$ 0.019); it is worth mentioning the moderate cut-off point variation among mentioned studies and the present [22]. In the present report, this OS tendency is evidenced although it does not reach statistical significance (Table 3).

P21 ${}^{\text{CIP1}}$ and p27 ${}^{\text{KIP1}}$ are inhibitors of the CIP/KIP family, whose function is the inhibition of DNA replication and cell proliferation. The loss of expression of these proteins has been related to worse OSCC prognosis [3, 4]. Evidences are particularly stronger towards p27 ${}^{\text{KIP1}}$ expression; Gao et al. performed a pooled analysis on its prognosis significance verifying that p27 ${}^{\text{KIP1}}$ low expression was significantly correlated with poorer OS for OSCC patients (HR $=$ 0.743 (95% CI 0.62–0.89, $p=$ 0.002) [18]. Evidences toward the effect of P21 ${}^{\text{CIP1}}$ in OSCC prognosis is much more diffuse as evidenced by our group on a comprehensive review [10]. Choi et al. obtained an inverse correlation tendency between the two markers ( $p=$ 0.08), although this group failed to observe any relation with clinical-pathological parameters [23]. Zhang et al. also obtained a statistically significant correlation between the expression of both CPs through Pearson analysis [24]. Fillies et al. did not observe any significant result regarding OS through Cox regression models [5]. Kapranos et al. analysing tumours of the oral cavity and larynx at the same time, found no correlation on their inmunoexpression, nor obtained differences in survival in any combinations of expression [25].

A bulk of research has ascertained the pivotal role of cyclins in oral carcinogenesis, specially the role of Cyclin D1. Cyclin D1 turnover and nuclear accumulation is regulated by NF- $\kappa$ B pathway activation and the inactivation/phosphorylation of GSK3b. A recent meta-analysis involving 2942 patients correlated its overexpression with worse OS for OSCC patients (HR $=$ 2.00, 95% CI $=$ 1.59–2.51, $p<$ 0.001) [17]. The main underlying mechanism explaining this rationale is amplification of the CCND1 gene, which in turns was verified in the present study via TCGA database analysis (Fig. 3) [15]. Nonetheless, other interferers can modulate its transcription; of which the best characterised are AP-1, ETS1, $\beta$ -Catenin, c-Myc, ETS1, and STAT3/5A [17]. Pignataro et al. studied the correlation of p27 ${}^{\text{KIP1}}$ and Cyclin D1 in head and neck squamous cell carcinoma finding it inverse-the majority of tumours that show positive expression of p27 ${}^{\text{KIP1}}$ , show negative expression of Cyclin D1 and tumours with positive expression of Cyclin D1, present negative expression of p27Kip1. This group also noticed that Cyclin D1 $+$ / ${}^{\text{KIP1}}$ tumours have a worse prognosis and that Cyclin D1-/p27 ${}^{\text{KIP1}}+$ in terms of OS ( $p=$ 0.0015) and recurrence-free survival ( $p=$ 0.0001) [26]. In this vein, Rodolico et al. also observed that tumours with overexpression of Cyclin D1 and with subexpression of p27 ${}^{\text{KIP1}}$ are related to the appearance of lymph node metastases in patients with squamous cell carcinomas of the lower lip [27].

Finally, authors would like to point out a few limitations. The discordance with previously published studies may be a result of geographic variations and ethnic disparity that result in different etiological factors for OSCC and also due to differences related to technical procedures such as the use of different monoclonal antibodies, cut-off points for positivity interpretation, or the use of univariate vs. multivariate analysis.

Overall, the present report ascertained neither the presence nor the degree of CPs expression influenced OS in the present OSCC cohort. Integration with TCGA database supported the presence of moderate cell cycle variations at a phenotype and genotype level in OSCC.