FormalPara Key Summary Points

The study compared the post-surgical treatment outcomes of infranotch T4b (IN–T4b) oral cancers with T4a oral cancers.

A systematic search was conducted of PubMed, EMBASE and Cochrane databases from 2000 to 2022. We included clinical studies reporting surgical and adjuvant therapy outcomes for patients with either IN–T4b or T4a tumours. The meta-analysis included 11,381 patients from 16 studies.

The pooled 2 year and 5 year overall survival (OS), disease-free survival (DFS), disease-specific survival (DSS) and locoregional control (LC) rates for IN–T4b patients were determined. There was no statistically significant difference in outcomes between IN–T4b and T4a oral cancers for these measures.

The study concluded that IN–T4b oral cancers may have similar outcomes to T4a oral cancers. This suggests the possibility of down-staging IN–T4b cancers to T4a and further stratifying T4a cancers based on anatomical tumour spread, moving IN–T4b out of the current T4b stage group that also contains patients with inoperable carotid artery encasement, skull-base or pterygoid plate involvement.

The findings may have important implications for the management and treatment decision-making process in oral cancer patients, suggesting reconsidering a subgroup of now ‘inoperable patients’ for surgery.

Introduction

Oral cavity cancer poses a significant burden worldwide. According to the latest Globocan data, 377,713 new oral cancer patients are diagnosed yearly [1]. The incidence of oral cancer is notably higher in South Asia, with approximately 70% of patients being reported in the locally advanced stages [2]. T4b oral cancers comprise tumours with either masticator space involvement, encasement of the carotid artery, skull-base or pterygoid plate involvement. T4a is defined as oral cancer involving the adjacent skin and/or bone. According to the 2023 National Comprehensive Cancer Network (NCCN) guidelines, the management of T4b oral cancers is combined along with the management of unresectable nodal disease and those unfit for surgery [3]. The NCCN guidelines suggest treatment with radiation therapy (RT) with or without chemotherapy (CT), and the intent of treatment is based on the patient’s performance status. In 2006 and 2007, Liao and colleagues from Taiwan showed that selected patients with masticator space involvement had reasonable survival outcomes following surgery and adjuvant RT/chemoradiation (CRT) [4, 5]. Outcomes in those with T4b disease (due to masticator space involvement) were comparable with T4a cancers in the cohort. This seminal article led to changes in the American Joint Committee on Cancer (AJCC) oral cavity cancer staging [6]. Furthermore, Liao et al. proposed further compartmentalisation of the infratemporal fossa (ITF) at the level of the mandibular notch–into infranotch and supranotch, based on the masticator-space extension remaining inferior to the level of the mandibular notch, or extending superiorly to it in T4b oral cancers. Surprisingly, in their series, the subset of infranotch T4b (IN–T4b) patients outlived T4a oral cancers [5]. Following their description of surgery for oral cancer with masticator space involvement, there has been an increase in the number of publications suggesting equivalent surgical outcomes for IN–T4b compared with T4a oral cavity cancer [7,8,9].

Here, we conduct a systematic review and meta-analysis to determine the overall curative intent surgical outcomes for IN–T4b oral cancers and to compare the outcomes with those of T4a patients.

Methodology

Search Strategy

Based on the assessing the methodological quality of systematic reviews (AMSTAR) 2 guidelines, at least two databases must be included in the search strategy for adequate literature coverage [10]. PubMed, EMBASE and Cochrane databases were searched to obtain comprehensive coverage of the published medical literature. Only published data in English language between 2000 and 2022 were considered.

Search Syntax

The following terms were included: ‘T4b’, ‘T4a’, ‘Oral cavity’, ‘Carcinoma’, ‘Cancer’, ‘Retromolar trigone’, ‘Buccal mucosa’, ‘Alveolus’, ‘Infratemporal fossa’, ‘Masticator space’, ‘Outcomes’. Boolean operators (NOT, AND, OR) were used to obtain the results. The last retrieval was on 9 March 2023.

Data Screening and Selection

The retrieved articles were screened independently by two investigators, KNR and RDA, based on the type of article, title and abstract. The eligible articles were pooled, and thorough full-text analysis and references in the relevant articles were further assessed by snowball searching (Fig. 1). KNR and RDA selected the articles, and the senior authors resolved any disagreement on including articles.

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Inclusion criteria

  1. 1.

    Treatment-naive T4a and T4b oral cancer (with masticator space or ITF involvement)

  2. 2.

    Original research articles published in peer-reviewed journals

  3. 3.

    The study must report at least one outcome following curative intent surgery – OS, DFS, DSS or LC

Exclusion criteria

  1. 1.

    Non-human studies

  2. 2.

    Neoadjuvant chemotherapy/radiation therapy

  3. 3.

    Any previous oncological treatment

  4. 4.

    Recurrent or second primary tumours

  5. 5.

    Not reported – re-operative outcomes

  6. 6.

    Non-original research reports, including review articles, meeting abstracts, case reports and editorial letters.

  7. 7.

    Incomplete data or insufficient information

  8. 8.

    Overlapping study populations shared dataset

Data Extraction

All included articles were independently screened by KNR and RDA. The following study characteristics were recorded: author, year of publication, country, sample size, type of study, level of evidence, Newcastle–Ottawa scale (NOS), risk of bias, number of T4b patients, number of T4a patients, 2 year and 5 year overall survival (OS), disease-free survival (DFS), disease-specific survival (DSS) and local control (LC) of the included studies were recorded and compiled (Table 1). Definitions: overall survival: the time from treatment completion to death due to any cause; disease-free survival: the time from treatment completion to disease recurrence or death; disease-specific survival: the time from treatment completion to death due to disease; local control: no local progression at the primary tumour site at specific temporal intervals [11].

Table 1 Overview of included studies
Full size table

Quality Assessment

Level of Evidence

The level of evidence of the eligible studies was performed independently by two authors (KNR and RDA), as per the Oxford Centre for Evidence-Based Medicine (OCEBM) criteria [12].

Methodology Quality

Two authors assessed methodological quality; the Newcastle–Ottawa scale was used to evaluate the quality of the included studies [13]. The score ranged from 0 to 9. The articles with a score > 5 were selected for systematic review.

Risk of Bias Assessment

The risk of bias assessment tool for the non-randomised trial tool from AMSTAR guidelines was used to determine the bias [14]. The following domains were assessed: patient selection, confounding variables, measurement of outcomes, incomplete data and selective reporting. The studies were graded as low risk, unclear risk and high risk of bias using RevMan v.5.4 (Cochrane Collaboration, Copenhagen, Denmark) (Supplementary Fig. 1).

Statistical Analysis

The meta-analysis applied the log odds ratio (logOR) to evaluate variations in outcome measures [15]. A random-effects model (REM) was chosen, considering both within-study and between-study variability [16]. This model accommodates differences in effect size across studies, making it more cautious than a fixed-effects model. Effect-size estimates were visually represented through Forest [17] and L’Abbé plots [18]. The nature of data distribution was assessed via Normal Quantile–Quantile (Q–Q) plots [19].

To evaluate heterogeneity, the DerSimonian–Laird (DL) estimator was utilised to approximate the between-study variance (tau-squared), which captures variability in true effect sizes among studies. The Cochran Q-test for heterogeneity was deployed to establish whether significant heterogeneity existed between individual study effect sizes and the overall effect size. Higgin’s I2 statistic, expressed as a percentage, conveyed the proportion of total variability across studies attributed to heterogeneity rather than chance. A value of 0% indicates no observed heterogeneity, while higher values indicate escalating heterogeneity levels [20]. In patients of detected heterogeneity (tau2 > 0), a prediction interval was offered to account for uncertainty stemming from heterogeneity. This interval, wider than a confidence interval, provided a range in which the true effect size of a new study is likely to fall [21]. Outliers were explored using studentised residuals, considering both estimated effect size variability and their standard errors [22]. Studies with studentised residuals surpassing a specific threshold (calculated with the Bonferroni correction) were identified as potential outliers and visualised through a radial plot [23]. Cook’s distances were employed to identify influential studies, quantifying each study’s impact on the overall model. Studies with Cook’s distances exceeding a particular threshold (median plus six times the interquartile range of Cook’s distances) were labelled as influential [24].

Furthermore, funnel plot asymmetry was assessed to detect potential publication bias, where smaller studies with non-significant results might be underrepresented. Rank correlation and regression tests were employed to examine funnel plot asymmetry using the standard error of observed outcomes as the predictor [25].

To derive the outcomes from the presented Kaplan–Meier plots, we employed a method involving measuring data points on the x axis, specifically at the 2 year and 5 year timepoints [26]. This process was facilitated by digitally recalibrating the provided Kaplan–Meier graphs using the Webplotdigitizer platform [27, 28]. By doing so, we could accurately determine the corresponding coordinates along the x and y axes.

All computations were executed using the R Project for Statistical Computing version 4.3.1 for Windows [29].

Reporting and Registration

This work has been reported concordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and assessing the methodological quality of systematic reviews (AMSTAR) guidelines. The meta-analysis was critically appraised by AMSTAR and was found to be a high-quality review. International prospective register of systematic reviews (PROSPERO) registration no. CRD42023437368 [30]. Ethical approval was not applicable as this article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Results

Literature Retrieval and Data Extraction

The initial literature search using search syntax identified 140 articles. Of these, 74 remained after deleting 66 duplicates. Upon title and abstract screening, 38 articles were omitted as they did not meet the inclusion criteria. After full-text analysis of the remaining 36 articles, 20 papers were excluded due to the non-availability of full text (n = 2), only published abstract (n = 1), previous cancer-directed treatment (n = 6), no survival data available for analysis (n = 5) and no data available for a primary surgical subset (n = 6). A total of 16 studies [5, 7,8,9, 31,32,33,34,35,36,37,38,39,40,41,42] were included in the qualitative synthesis. Finally, seven studies [7, 8, 31, 33, 37, 39, 41] were included for meta-analysis as T4a and T4b data were available for comparison. The data were extracted from the published Kaplan–Meier graphs of the respective articles using Webplotdigitizer.

The Rationale for Study Inclusion

Liao et al. [31] (2012) provided distinct datasets for infranotch T4b and T4a categories within their findings. Pillai et al. [8] have categorised T4b tumours into three distinct classes (I: lower masticatory space or infranotch; II: intermediate masticatory space or low supra-notch; and III: high-masticatory space or high supranotch) based on the extent of infiltration into the ITF, aligning with the classification outlined by Trivedi et al. [43]. In our meta-analysis, we have selectively incorporated the subset of T4b patients falling under class I from this study. Similarly, Pillai et al. and Mair et al. [33] have exclusively examined patients with low ITF and excluded those involving supranotch ITF as per the classification elucidated by Liao et al. [5] (2007). Thiagarajan et al. [37] recently conducted a comparative investigation of upfront surgery versus neo-adjuvant chemotherapy for both T4a and T4b oral cancers. Within their upfront surgery group, they have encompassed patients that underwent curative-intent surgery. Our analysis has confined our focus to the cohort of upfront surgical cases encompassing T4a and a subset of T4b. Kumar et al. [39], in their study, have included patients that underwent curative-intent surgery for T4b. Their scrutiny of T4b outcomes has been predicated on the involvement of components within the masticator space, including the masseter muscle, medial pterygoid muscle, ramus of mandible, lateral pterygoid muscle and the lower pterygoid plate. Remarkably, the components scrutinised by Kumar et al. [39] closely correspond to the constituents of low ITF, as delineated in the subcategorisation framework proposed by Liao et al. [5] (2007).

Further, Patel et al. [7] have incorporated data from the NCDB database, while Kang et al. [41] have integrated information from the Taiwanese Cancer Registry. An important note is that Patel et al. [7] study exhibits disparities in the numerical values depicted within their Forest plot. In this analysis, we have confined our consideration to values concordant with the data described within the results section and consistent with their abstract. Both Kang et al. [41] and Patel et al. [7] have encompassed patients who underwent curative-intent surgical interventions for T4a and T4b. However, neither of these studies provided distinct data for patients falling within the supranotch and infranotch T4b categories.

Quality of Included Studies

The main characteristics of the included studies are summarised in Table 1. The qualitative data synthesis yielded 1229 surgically treated T4b patients from 16 articles, and 820 patients from 7 studies were included in the final meta-analysis. Seven studies reported the comparative surgical outcomes for T4a and T4b patients; 10,561 patients of surgically treated T4a patients were included for comparative meta-analysis. Eligible studies were either a prospective (n = 1) [8] or retrospective (n = 15) [5, 7, 9, 31,32,33,34,35,36,37,38,39,40,41,42] cohort study design, and none were randomised controlled trials. Two studies were based on the cancer registry, namely the Taiwan cancer registry (TCR) and the US National Cancer Database (NCDB). The Newcastle–Ottawa score ranged from 6 to 8 (Table 1). Based on the risk of bias assessment tool, the included studies had the highest risk of patient selection and confounding variables (Supplementary Fig. 1).

Two Year Overall Survival (T4b versus T4a)

Two year OS data were extracted from 11 studies [7, 31,32,33,34,35, 37, 38, 40,41,42] for T4b disease. The 2 year OS data were extracted from five studies [7, 31, 33, 37, 41] for T4a and T4b. The pooled 2 year OS was 61.2% for T4b patients from 11 studies [7, 31,32,33,34,35, 37, 38, 40,41,42]. The pooled 2 year OS was 59.7% for T4b (n = 660) and 64.6% for T4a (n = 10,371) patients from five studies in the meta-analysis [7, 31, 33, 37, 41]. The pooled data had significant heterogeneity (Table 2). The logOR was 0.21 (−0.25, 0.67; DL, REM, CI = 95%) favouring T4a and the overall effect estimates for T4a and T4b were not statistically significant for 2 year OS (Z = 0.9, p = 0.36, CI = 95%) (Fig. 2). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was insignificant (Table 2) (Supplementary Figs. 2, 3, 4).

Table 2 Overview of meta-analysis statistics
Full size table
Fig. 2
figure 2

Forest plot (A, B) and L'Abbé plot (C, D) for overall survival (OS); 2 year OS (A, C) and B 5 year OS (B, D)

Full size image

Two Year Overall Survival (subset IN–T4b versus T4a)

Two year OS data were extracted from three studies [31, 33, 37] for T4a and IN–T4b patients for the meta-analysis. The pooled 2 year OS was 59.3% for IN–T4b (n = 149) and 65.3% for T4a (n = 653) patients from three studies in the meta-analysis [31, 33, 37]. The pooled data had significant heterogeneity (Supplementary Table).The logOR was 0.28 (−0.47, 1.03; DL, REM, CI = 95%) favouring T4a and the overall effect estimates for T4a and T4b were not statistically significant for 2 year OS (Z = 0.73, p = 0.46, CI = 95%). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was significant (Supplementary Figs. 5–9).

Five Year Overall Survival (T4b versus T4a)

Five year OS data were extracted from nine studies for T4b disease [5, 9, 31, 33, 35, 38, 39, 41, 42]. Five year OS data were extracted from four studies [31, 33, 39, 41] for T4a and T4b patients for the meta-analysis. The pooled 5 year OS was 46.1% for T4b patients from nine studies [5, 9, 31, 33, 35, 38, 39, 41, 42]. The pooled 5 year OS was 53.2% for T4b (n = 508) and 50.9% for T4a (n = 4393) patients from four studies [31, 33, 39, 41] in the meta-analysis. The pooled data had significant heterogeneity (Table 2). The logOR was 0.09 (−0.20, 0.38; DL, REM, CI = 95%) favouring T4a and the overall effect estimates for T4a and T4b were not statistically significant for 5 year OS (Z = 0.61, p = 0.54, CI = 95%) (Fig. 2). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 10–12).

Five Year Overall Survival (subset IN–T4b versus T4a)

Five year OS data were extracted from three studies [31, 33, 39] for T4a and IN–T4b patients for the meta-analysis. The pooled 5 year OS was 53.3% for IN–T4b (n = 153) and 49.6% for T4a (n = 362) patients from three studies in the meta-analysis [31, 33, 39]. The pooled data had significant heterogeneity (Supplementary Table). The logOR was 0.7 (−0.4, 1.8; DL, REM, CI = 95%) favouring T4a and the overall effect estimates for T4a and IN–T4b were not statistically significant for 5 year OS (Z = 1.24, p = 0.21, CI = 95%). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was significant (Supplementary Figs. 13–17).

Two Year Disease-Free Survival (IN–T4b versus T4a)

Two year DFS data were extracted from nine studies [5, 8, 31, 33, 35,36,37,38, 40] for IN–T4b disease. Two year DFS data were extracted from four studies [8, 31, 33, 37] for T4a and IN–T4b patients for the meta-analysis. The pooled 2 year DFS was 57.9% for IN–T4b patients from nine studies [5, 8, 31, 33, 35,36,37,38, 40]. The pooled 2 year DFS was 60% for IN–T4b (n = 279) and 63.2% for T4a (n = 749) patients from four studies [8, 31, 33, 37] in the meta-analysis. The pooled data had significant heterogeneity (Table 2). The logOR was 0.22 (−0.35, 0.79; DL, REM, CI = 95%), favouring T4a patients and the overall effect estimates for T4a and IN–T4b were not statistically significant for 2 year DFS (Z = 0.74, p = 0.45, CI = 95%) (Fig. 3). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 18–20).

Fig. 3
figure 3

Forest plot (A, B) and L'Abbé plot (C, D) for disease-free survival (DFS); 2 year DFS (A, C) and B 5 year DFS (B, D)

Full size image

Five Year Disease-Free Survival (IN–T4b versus T4a)

Five year DFS data were extracted from nine studies for IN–T4b disease [5, 8, 9, 31, 33, 35, 36, 38, 39]. Five year DFS data were extracted from four studies for T4a and IN–T4b patients [8, 31, 33, 39] for meta-analysis. The pooled 5 year DFS was 48.4% for IN–T4b patients from nine studies [5, 8, 9, 31, 33, 35, 36, 38, 39]. The pooled 5 year DFS was 51.6% for IN–T4b (n = 283) and 55.9% for T4a (n = 458) patients from four studies in the meta-analysis [8, 31, 33, 39]. The pooled data had significant heterogeneity (Table 2). The logOR was 0.17 (−0.42, 0.77; DL, REM, CI = 95%), favouring T4a patients but the overall effect estimates for T4a and IN–T4b were not statistically significant for 5 year DFS (Z = 0.57, p = 0.57, CI = 95%) (Fig. 3). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 21–23).

Two Year Disease-Specific Survival (T4b versus T4a)

Two year DSS data were extracted from three studies for T4b disease [31, 38, 41]. Two year DSS data were extracted from two studies for meta-analysis for T4a and T4b patients [31, 41]. The pooled 2 year DSS was 66.1% for T4b patients from three studies [31, 38, 41]. The pooled 2 year DSS was 67.5% for T4b (n = 403) and 71.5% for T4a (n = 4164) patients from two studies in the meta-analysis [31, 41]. The pooled data had significant heterogeneity (Table 2). The logOR was 0.17 (−0.55, 1.02; DL, REM, CI = 95%), slightly favouring T4a but the overall effect estimates for T4a and T4b were not statistically significant for 2 year DSS (Z = 0.58, p = 0.56, CI = 95%) (Fig. 4). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 24–26).

Fig. 4
figure 4

Forest plot (A, B) and L'Abbé plot (C, D) for disease-specific survival (DSS); 2 year DSS (A, C) and B 5 year DSS (B, D)

Full size image

Two Year Disease-Specific Survival (IN–T4b versus T4a)

Two year DSS data for T4a and IN–T4b patients were reported in Liao et al. [31]. The 2 year DSS was 72% for IN–T4b (n = 48) and 68% for T4a (n = 133) patients. Unfortunately, a single study cannot be meta-analysed.

Five Year Disease-Specific Survival (T4b versus T4a)

Five year DSS data were extracted from three studies for T4b disease [31, 38, 41]. Five year DSS data were extracted from two studies for meta-analysis for T4a and T4b patients [31, 41]. The pooled 5 year DSS were 58.3% for T4b patients from three studies [31, 38, 41]. The pooled 5 year DSS were 61.5% for T4b (n = 403) and 62% for T4a (n = 4164) patients from two studies in the meta-analysis [31, 41]. The pooled data had significant heterogeneity (Table 2). The logOR was 0.08 (−0.65, 0.8; DL, REM, CI = 95%), favouring T4a patients but the overall effect estimates for T4a and T4b were not statistically significant for 5 year DSS (Z = 0.21, p = 0.83, CI = 95%) (Fig. 4). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 27–29).

Five Year Disease-specific survival (IN–T4b versus T4a)

Five year DSS data for T4a and IN–T4b patients were reported in Liao et al. [31]. The 5 year DSS was 68% for IN–T4b (n = 48) and 60% for T4a (n = 133) patients. Unfortunately, a single study cannot be meta-analysed.

Two Year Local Control (IN–T4b versus T4a)

Two year LC data were extracted from three studies for IN–T4b disease [8, 33, 35]. Two year LC data were extracted from two studies [8, 33] for T4a and IN–T4b patients for meta-analysis. The pooled 2 year LC was 47% for IN–T4b patients from two studies [33, 35]. The 2 year LC was 44.5% for IN–T4b (n = 205) and 56% for T4a (n = 231) patients from two studies [33, 35]. The pooled data had significant heterogeneity (Table 2). The logOR was 0.47 (−0.33, 1.26; DL, REM, CI = 95%), favouring T4a patients and the overall effect estimates for T4a and IN–T4b were not statistically significant for 2 year LC (Z = 0.47, p = 0.25, CI = 95%) (Fig. 5). Normal Q–Q plot shows bimodal distribution. The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 30–32).

Fig. 5
figure 5

Forest plot (A, B) and L'Abbé plot (C, D) for local control (LC); 2 year LC (A, C) and B 5 year LC (B, D)

Full size image

Five Year Local Control (IN–T4b versus T4a)

Five year LC data were extracted from three studies for IN–T4b disease [31, 33, 35]. Five year LC data were extracted from two studies for T4a and IN–T4b patients for meta-analysis [31, 33]. The pooled 5 year LC was 47% for IN–T4b patients from three studies [31, 33, 35]. The 5 year LC was 56% for IN–T4b (n = 123) and 62.5% for T4a (n = 268) patients from two studies [31, 33] in the meta-analysis. The pooled data had moderate heterogeneity (Table 2). The logOR was 0.34 (−0.31, 0.99; DL, REM, CI = 95%), favouring T4a patients but the overall effect estimates for T4a and IN–T4b were not statistically significant for 5 year LC (Z = 1.01, p = 0.31, CI = 95%) (Fig. 5). Normal Q–Q plot shows bimodal distribution (Supplementary Material). The publication bias (funnel plot asymmetry) was not significant (Table 2) (Supplementary Figs. 33–35).

Citation Network Analysis

The Litmaps tool was used to generate the citation network among the articles included in the study [44]. The literature citation maps display axes on a logarithmic scale of citations distributed across the article positions based on the semantic similarity of their titles. The size of each article bubble corresponds to the logarithmic scale of its citation count. Notably, the citation network indicated that the work by Liao et al. [5] had a more extensive network of citations, followed by Mair et al. [33]. Interestingly, almost all the included studies referenced Liao et al. [5], followed by citations to Mair et al. [33] It is worth mentioning that the Thiagarajan et al. [37] study was unique in that it did not cite any of the other included studies (Fig. 6).

Fig. 6
figure 6

Citation network

Full size image

Discussion

Managing oral cancers is still challenging, even for experienced multidisciplinary centres. Surgery is the cornerstone in the management of oral cavity cancers. The masticator space is a suprahyoid musculofascial compartment consisting of mastication muscles, the mandible ramus, a neurovascular bundle and fibrofatty tissue [45]. Liao et al. first described the outcome after surgery for T4b cancers. They subdivided T4b (due to masticator space involvement) cancers into infranotch and supranotch disease, depending on the cranial extent in the masticator space remaining below or extending above the mandibular notch below the mandibular notch on cross-sectional imaging [5]. Trivedi and colleagues described compartmental clearance for supranotch T4b tumours invading the ITF [46]. Despite a well-defined surgical technique for managing IN–T4b cancers, it remains elusive among many surgeons due to the risk of severe peri-operative complications [43]. Also, it is regarded as an unresectable disease, or there is a high risk of positive surgical margins [47]. Today, composite en-bloc resection for IN–T4b oral cancers is routinely performed in many high-volume treatment centres and is considered a standard of care. Nevertheless, as with all surgeries, especially rare, new and demanding techniques, standardisation of surgeries within the same groups is challenging [48].

In this study, we have restricted our analysis to compare the surgical outcomes of IN–T4b with T4a cancers, as there is minimal evidence to support surgery for supranotch T4b over non-surgical treatment [5]. The inferior results of surgery in supranotch T4b disease are mainly due to the inability to achieve R0 clearance, higher tumour margin positivity rates and complications following surgery [33, 47].

In this study, the 2 year OS rate among T4a patients was marginally higher in comparison with IN–T4b disease (65.3% versus 59.3%), with logOR of only 0.28 ( −0.47, 1.03) (p = 0.46). Surprisingly, Liao et al. [31] showed a better 2 year OS in IN–T4b compared with T4a oral cancers. The 5 year OS [T4a 49.6%, IN–T4b 53.3%, logOR 0.7 (−0.4, 1.8), p = 0.21], 2 year DFS [T4a 63.2%, IN–T4b 57.9%, logOR 0.22 (−0.35, 0.79), p = 0.45], 5 year DFS [T4a 55.9%, IN–T4b 51.6%, logOR 0.17 (−0.42, 0.77), p = 0.57], 2 year LC [T4a 56%, IN–T4b 44.5%, logOR 0.47 (−0.33, 1.26), p = 0.25] and 5 year LC [T4a 62.5%, IN–T4b 56%, logOR 0.34 (−0.31, 0.99), p = 0.31] of IN–T4b cancers was not statistically different from T4a cancers. Several authors have reported better outcomes (2 year OS, 5 year OS, 2 year DFS, 5 year DFS, 2 year DSS, 5 year DSS and 5 year LC) in patients with T4b disease compared with those with T4a tumours [8, 31, 39].

The primary evidence for 2 year DSS and 5 year DSS with surgery for IN–T4b disease was derived from Liao et al. [31], showing a 2 year DSS of 68% for T4a and 72% for IN–T4b and the 5 year DSS of 60% for T4a and 68% for IN–T4b. While the overall effect was not statistically significant, the results of the studies are suggestive of slightly higher rates of recurrence in T4b, in which increased tumour margin positivity rates may explain subsequent surgery for IN–T4b disease. This warrants a need for further optimisation of surgical techniques for IN–T4b disease. Furthermore, an inherent selection bias in selecting the IN–T4b patients for surgery in these series must be acknowledged when interpreting these numbers.

Liao et al. [5] reported a close margin (< 4 mm) rate of 13.2% for T4b patients in 2006. In their 2012 study, the same authors [31] found a close margin rate of 13% for T4a patients and 14.6% for T4b patients (p = 0.85). Trivedi et al. [32] 2015 observed a tumour margin positivity rate of 6.7%, and their findings indicated that a positive margin near the pterygomaxillary fissure was associated with a high risk of local failure and potential intracranial extension. Mair et al. [33] documented close margin (< 5 mm) rates of 7.4% for T4a patients and 12% for T4b patients (p = 0.3). Mohiyuddin et al. [34] found a close margin (< 5 mm) rate of 25% in their T4b cohort. In the study by Pillai et al. [8], tumour margin positivity rates were 0.9% for T4a patients and 6.8% for T4b patients. Katna et al. [35] reported a tumour margin positivity rate of 13% for T4b cancer, primarily observed in the ITF soft tissue. Brinda et al. [36] noted a close margin (< 5 mm) rate of 40% in their T4b cohort. In the study by Thiagarajan et al. [37], the close margin (< 5 mm) rate was 10% for T4b patients undergoing upfront surgery. Surprisingly, the study by Baddour et al. [38] revealed a high close margin (< 5 mm) rate of 96% in their T4b patients. Kumar et al. [39] found an overall close margin (< 5 mm) rate of 9.8%. In the Patel et al. [7] study, positive margins were identified in 31.4% of patients. Close margins (< 5 mm) were identified in 37.3% of patients in the study by Gangopadhyay et al. [40] Lastly, Lin et al. [42] observed positive margins (< 1 mm) in 34.6% of patients.

Also, there is a trend towards higher rates of lymphovascular invasion, perineural invasion, tumour budding and extranodal extension in T4b patients. The subset of IN–T4b cancers in Liao et al. [4] had better outcomes than T4a cancers mainly due to a higher nodal disease burden in their T4a cohort, although this was not reported in other studies. Some centres have tried to propose neoadjuvant chemotherapy (NACT) as an induction treatment for T4b patients with varying results [8, 43]. We cannot comment on the role of NACT as it was an exclusion criterion for studies to be included in our systematic review and meta-analysis.

Many classifications have been proposed for the compartmentalisation of the ITF based on the extent of disease on a coronal section of contrast-enhanced computerised tomography [5, 49]. The ‘supra’-mandibular ‘notch’ and ‘infra’-mandibular ‘notch’ concept initially described by Liao et al. [5] in 2006 is often used. The classification provided by Mahajan et al. [49] still needs to be externally validated. Similarly, Trivedi et al. have also described the subcompartmentalisation of ITF based on the structures involved [43].

With advances in imaging technologies, we should aim to further subcategorise the individual component of T4a and T4b according to anatomical involvement. Buccal mucosa and retromolar trigone cancers typically invade the masseter or medial pterygoid muscle early in the disease process as it is generally close to the invasive front of the tumour. To better understand tumour spread, we should uniformly determine and record the reasons for categorising the disease as T4a or T4b; i.e., a tumour invading only skin, bone, masseter muscle, medial pterygoid muscle and a combination of these must be meticulously identified and reported. This may be cumbersome initially but, prospectively, it will provide valuable insight into the disease process.

The operable IN–T4b disease is currently included with inoperable tumours with encasement of the carotid artery, skull base and/or pterygoid plate involvement. These inoperable tumours have vastly different outcomes compared with operable tumours. Our meta-analysis is the best evidence suggesting equivalent survival between T4a and IN–T4b oral cancers. Hence, we propose reclassifying current IN–T4b patients to T4a. The newly proposed T4a categorisation must be further substratified according to the anatomical structure involved (skin, bone, masseter muscle, medial pterygoid muscle and combination). A similar type of substratification of the AJCC staging system exists for cutaneous malignant melanoma.

Strengths: This is the first meta-analysis comprehensively comparing the surgical outcomes in infranotch masticator space T4b and T4a oral cancers, with a comprehensive summary of the available evidence.

Limitations: Observing that our systematic review and meta-analysis mainly included retrospective cohort studies, the likelihood of selecting favourable IN–T4b tumours for surgical treatment in all these studies is high, and the selection criteria used probably differ among included studies. The latter impedes applying the summary data from the current systematic review and meta-analysis to daily practice since uniform and unequivocal selection criteria are required. This forms a research area for the future. Acknowledging the inherent limitations tied to database analyses, encompassing censored data and instances where patients are lost to follow-up, is imperative. The nodal metastasis, adverse features, surgical techniques, functional outcomes and adjuvant therapies were not analysed as no data for comparing T4a and IN–T4b disease were reported.

Conclusions

In this meta-analysis, the surgical outcomes for IN–T4b cancer are similar to those of T4a oral cancers. Consideration should be given towards reclassifying the current IN–T4b cancers to T4a as they are currently staged with inoperable carotid artery encasement, skull-base or pterygoid plate involvement. We also recommend further stratification of the new T4a category based on anatomical tumour spread.