Tumor associated tissue eosinophilia in oral squamous cell carcinoma: A systematic review and meta-analysis

Abstract

Background

Oral squamous cell carcinoma (OSCC) still lacks the valid prognosticator for survival. Tumor associated tissue eosinophilia (TATE) have been widely investigated in the literature but still there is no consensus viewpoint on its role as prognostic marker. With this view in mind, a systematic review and meta-analysis of studies that evaluated the overall risk of elevated TATE and survival in OSCC was conducted.

Methodology

A systematic literature search conducted to identify studies that have evaluated TATE in OSCC. Overall survival (OS), Disease Free Survival (DFS) Hazard ratio (HR) with 95% confidence interval) was extracted. HR was extrapolated if direct HR was not given in publication. Studies were pooled and combined hazard ratios (HRs) of High TATE for OS and DFS were calculated.

Results

6 out of 874 articles were found eligible for systematic review and meta-analysis. High TATE was predictor of OS (pooled HR 0.45 (95% CI, 030–0.65, P ​< ​0.0001)) and DFS (pooled HR 2.33 (95% CI, 0.74–7.37)) in OSCC patients.

Conclusion

High TATE positively predicts OS and DFS in OSCC and thus can be regarded as an important survival prognosticator. Standardization of counting and reporting methodology for TATE count is recommended on future cohort studies.

1. Introduction

Several histopathological prognostic markers have been widely investigated and suggested to assist surgeon in treatment planning and determining prognosis for oral squamous cell carcinoma (OSCC) patients.¹ Although they have been proved useful in individual studies, they still lacks strong validation in clinical practice.² The various histopathological prognostic markers are tumor size, histological grade, tumor microenvironment including tumor infiltrating cells or immune cells. Immunity has been getting traction as result of better understanding of immune system due to advancement modern technology. Various tumor infiltrating and immune cells are part of immune surveillance in the body tissue and has been described to influence development of OSSC and its progression.³ Although, these cells are part of tumor microenvironment (TME), eosinophils is prominently present in TME and is associated with prognostic outcome in various malignancies of the body.

Eosinophils are potent immune modulatory, cytotoxic cells present in peri-tumoral and intra-tumoral tissue resulting in tissue eosinophilia. Tumor associated tissue eosinophilia (TATE) is, basically, presence of eosinophils in tumor stroma in absence of any ulceration or necrosis. TATE as prognostic biomarker is recommended in various malignancies associated with different body parts.⁴ Furthermore, quantification of eosinophils in tumor tissue was proposed as part of the histological evaluation to predict the patient prognosis.⁴ With regard to OSCC, there are divergent reports regarding the potential aspect of TATE or eosinophils as biomarker for prediction of prognosis outcome. However, TATE association with prognosis in OSCC is still unclear. Similar to other tumor, the histological studies of OSCC have reported prominent presence of eosinophils in TME resulting in TATE. Considering that histopathology examination is standard practice among pathologist, pathologist can perform the quantitative evaluation of eosinophils in tumor tissue. Due to lack on consensus on the role of TATE in OSCC, we realized a dire need for meta-analysis to better understand it role.

With this view in mind, a systematic review and meta-analysis was conducted on the overall risk of elevated TATE and survival in OSCC. Efforts were also made to assess the extent of reporting concerning clinico-pathological data, special histochemical and immunohistochemical investigation etc. using REMARK guidelines.³

2. Materials and methods

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines for the systematic review and meta-analysis were followed to evaluate of TATE in of OSCC (Supplementary Table S1).

2.1. Research question(s)- the following questions were framed according to PICO framework of study design

Questions- (A) Does elevated eosinophils or eosinophilia (I) in tumor tissue indicate better prognosis and survival (Outcome) in OSCC patients (Participants).

• (B)Can TATE be used as prognostic factor/biomarker in OSCC.

2.2. PICOTS statement

Participants: Patients with OSCC, confirmed by histopathology, with no history of chemotherapy or radiotherapy prior to tissue biopsy and treated either or in combination of chemotherapy, radiotherapy, surgery. All TNM stage patients were considered.

Index prognostic factor: High (Intense/heavy) tumor associated tissue eosinophilia (as percentage or mean/median count per millimeter square in tumor tissue) as single biomarker.

Comparator prognostic factors: TNM staging, histological grade, Low (absent/mild/moderate) tumor associated tissue eosinophilia.

Outcome: All cause mortality, recurrence or remission of OSCC.

Timing: Eosinophil estimation in tumor tissue was carried out before any type of intervention. Moreover, other post-operative data (all time-periods) was also considered.

Setting: Eosinophil count or TATE studied hospital based educational institutional settings for providing diagnostic and prognostic services to the OSCC patients; healthcare professionals may use this information for treating and managing OSCC patients.

2.3. Key definitions

Overall survival (OS) was defined as “the time interval (months) between primary treatment and death by oral cancer or last follow-up”. Disease-free survival (DFS) was defined as “the time interval (months) between primary treatment and the first recurrence (whether local, regional or distant)”.

2.4. Selection criteria

The following inclusion and exclusion criterion was used for retrieving the appropriate studies from the literature:

Inclusion Criteria: (i) studies that analyzed TATE and the prognostic/survival outcome in primary OSCC, (ii) addressed the prognostic value of TATE by univariate and/or multivariate survival analyses of at least 10 OSCC patients, (iii) studies with univariate or multivariate analysis, (iv) studies that reported Hazard ratio (HR) or Odd ratio (OR) or risk ratio (RR) and 95% CI values, (v) studies with Kaplan–Meier (KM) curves with HR and 95% CI values from the KM curves, (vi) minimum 1-year follow-up period to note outcome (vii) articles in English or those having detailed summary in English and (viii) studies published till date with no restriction on time frame.

Exclusion Criteria: (i) Studies on cell lines and animal models, (ii) articles not in English language, (iii) abstracts, conference proceedings, letters to editors, editorial studies and (iv) original research papers on TATE in OSCC without information on tumor grading, TNM staging, TATE and outcome/Endpoint, (v) Tissues exposed to radiotherapy and/or chemotherapy were excluded.

2.5. Information sources and search strategy

Authors performed a comprehensive search of the databases such as PubMed, Medline, SCOPUS and Web of Science with selected keyword curated from Medical Subject Headings (MeSH) from PubMed in combination with Boolean operators AND (Supplementary Table S2). Additional search were performed with keyword obtained from retrieved articles. These articles were screened for addition references. Studies published till date with no restriction on time frame were screened. (Keywords included a combination were (“tumor associated tissue eosinophilia” OR “TATE” OR “tissue eosinophilia” OR “eosinophil count” OR “eosinophils”) AND (“oral squamous cell carcinoma” OR “oral cancer” OR “head and neck” OR “head and neck cancer”) AND (“TNM staging” OR “TNM stage” OR “histopathology” OR “histopathology grade/grading”) AND (“prognosis” OR “survival” OR “recurrence” OR “remission”)

2.6. Study screening and selection

Two reviewers (NC and GS) screened the title and abstract of retrieved article for the initial eligibility of study as per inclusion criteria. Two reviewers reviewed eligible articles full text independently and any discrepancy were sorted out with consensus.

2.7. Data collection

Before initiation of data collection, Reviewers (SS, SP, MY) discussed the data required for analysis to obtain the answers to questions and data collection form was created. To achieve the outcome of this systemic review and meat-analysis, two reviewers (MY and NC) collected following data items in pre-defined form. The data items included were: (i) author (Year)- name of the first author and year, (ii) country-geographic location of participants, (iii) sample/participants-number of samples investigated, (iv) age and gender, (v) tumor Site-location, (vi) TNM Staging, (vii) histological grade, (viii) eosinophil count (percentage or number per field)- mean, SD, (ix) clinical outcome and follow-up (alive, diseased, recurrence, metastasis)- status and follow-up duration, (x) HR for increased/elevated eosinophil count/TATE and (x) 5 years or 10 year overall survival (OS) rate or disease/progression free survival (DFS/PFS) rate, log hazard ratio (InHR), p value).

HR and their 95% CIs were retrieved from the original research articles and HR of more than one was regarded as poor outcome. Total numbers of deaths reported and number of patients in each category were considered in case HR information not available.⁵ Graphical survival plots were used to extract the data when only Kaplan-Meier curves was available. Otherwise, the estimation of the HR was performed by the described method.⁵

2.8. Quality and risk of bias (RoB) assessment

The quality of the studies were evaluated using the REporting recommendations for tumor MARKer prognostic studies (REMARK) criteria⁶ and methodology for scoring was adopted from Creemer et al. study.⁷ The two reviewers (Initials) assessed and scored the quality of included studies according to the REMARK guidelines whereas the QUIPS tool (Quality in prognostic factor studies) was used to examine each study’s RoB. GRADE methodology was utilized to derive level of evidence from summary this meta-analysis.⁸

2.9. Statistical analysis

The heterogeneity of the combined HRs was performed by Cochran’s Q test and Higgins I-squared statistics. P ​< ​0.05 was considered significant. The random effects model was castoff if heterogeneity was detected (P ​< ​0.05). The publication bias was evaluated with the funnel plot with the Egger’s bias indicator test (P values ​> ​0.05).⁹ Statistical analyses were performed by using the statistical software Revmen 5.3 and Microsoft Excel (Version 16).

3. Result

3.1. Search results

Initial search yield 874 records were retrieved from PubMed, Medline, SCOPUS and Web of Science. 161 were omitted after screening of the title and abstract for being review articles, letters, laboratory studies, studies with important data missing and studies inappropriate to the existing investigation. For example, some studies analyzed the predictive value of other factors such as tissue retraction interleukin-6, COX2, but not TATE. Some studies examined the TATE levels at one time point without follow-up to detect the outcome. One study was excluded because HR was missing or HR could not be calculated due non-availability of summary statistics. The final meta-analysis for TATE was performed for the remaining six studies (Fig. 1-PRISMA Flow chart).

Fig. 1.

Flow chart demonstrating process for identified studies in the searches along with reason for exclusion.

3.2. Study characteristics

The features of selected studies are abridged in Table 1. We collected the data from six studies, which elaborated 470 (Male 343; Female- 127) patients from the Brazil, Finland, India, and Sweden. In these six studies, TATE values were investigated by various means in each study. Two studies dealt with TATE as a dichotomous variable, with different cutoff values. Two studies dealt with TATE as a trichotomous variable and compared the results of the highest tertile with the lowest tertile. All the selected studies were retrospective and majority patient samples were of primary OSCC. Of all the selected studies, only one selected study presented HRs with 95% CI while in the other five studies, HRs were absent, and we needed to ask authors or calculate the HRs from the available data. The follow-up period in all the studies ranged from 0 to 287.4 months.

Table 1.

Baseline characteristics of included studies.

First author, date (ref.)	Study design	n	Age (years)	Gender	Cutoff/Catogory	Survival analysis	Hazard ratio	Follow-up, Months (range)	Hazard Ratio	Log Hazard Ratio	SE	Var(IN(HR))	Survival Outcome of high eosinophil cell count	log HR value direction	Tumour area	Direction of effect on survival of high eosinophil cell count
Dorta RG et al., 2002	R	125	30-95 (Mean ​- ​58)	M:105, F: 20	Mild: 0–26/mm2 Moderate: 27–83/mm2 Intense: >84/mm2	OS, DFS	Data extrapolated	(0–287.4)	OS: 0.506	OS: −0.68 DFS: −0.59	OS: −0.279 DFS: −0.26	OS-0.078; DFS-0.068	Better outcome	Negative	ITV, tumor proper	Positive
Oliveira D et al., 2009	R	43	28-83 (mean ​= ​55.79)	M:27 ​F:16	Mild: 0–175/mm2; Intense: 176–1317/mm2	DFS	Data extrapolated	(3–229)	DFS: 0.207	DFS: 1.573	DFS: 0.602	OS-; DFS: 0.3636	No outcome	Positive	ITV	No
Oliveira DT et al., 2012	R	71	35-77 (Median-59)	M:55 ​F:16	Absent⁄Mild (0–67eosinophils/mm2) Intense (68–298 eosinophils/mm2)	DFS	Data extrapolated	60	3.304	1.195	0.458	0.458	Poor Outcome	Positive	ITV	Negative
Jain D et al., 2017	R	87	47 ​± ​11 (25–70) Mean ​± ​Standard Deviation (Range)	M:76 F:11	Absent/Mild: 0–5.5/hpf Moderate: 6–42/hpf Intense: 43–300/hpf	OS	Report	(1.1–61.8)	0.281	−1.27	0.5	0.25	Better outcome	Negative	ITV, tumor proper	Positive
Rakesh N et al., 2015	R	45	45–76	M:25; F:20	Absent/mild: 0–10/hpf, Moderate: 10–30/hpf, Intense: >30/hpf	DFS	Data extrapolated	60	3.896	1.36	0.41	0.17	Poor Outcome	Positive	ITV	Negative
Peurala E et al., 2018	R	99	(Oral/Lip) 64.3 (36–89)/70.6 (27–87)	M:55 F:44	Less 1/PDF 1-29/HPF >30/HPF	OS	SC	40.7	0.454	−0.79	0.324	0.1052	Better outcome	Negative	ITV	Positive

Study design is described as consecutive patients (C), prospective (P), or retrospective (R). (−) ​= ​not reported; AP ​= ​author provided; BC ​= ​breast cancer; RFS ​= ​Recurrence free survival; DE ​= ​data extrapolated; DFS ​= ​disease-free survival; OS ​= ​overall survival; SC ​= ​survival curve. HPF ​= ​High power field. IF ​= ​Impact factor.

3.3. Overall survival and disease specific survival

For OS, a fixed effect model was applied since the p value of between-study heterogeneity was 0.59. As illustrated in Fig. 2A, increased/intense TATE significantly predicted better OS outcome, with a combined HR of 0.45 (95% CI, 030–0.65, P ​< ​0.0001). For DFS, a random effects model was applied. The P values of between-study heterogeneity were P ​< ​0.0001 for DFS analyses. As illustrated in Fig. 2B, the combined HR of 2.33 (95% CI, 0.74–7.37) showed that the increased eosinophils/intense TATE level had no significant relationship with DFS in OSCC patient. From the above data, TATE proved to be a prognostic biomarker for OS and DFS in OSCC. High TATE is associated with better prognosis in OSCC in term of survival; however, there is more risk of recurrence/metastasis in OSCC patients.

Fig. 2.

Forest plot- Forrest plots of studies evaluating hazard ratios (HR) with 95% confidence interval (95% CI) for high levels eosinophils/TATE as compared with low levels. Survival data are reported as overall survival (A), disease-free survival (B).

3.4. Quality assessment

Evaluation of the study quality using the REMARK criteria resulted in a mean quality of 6.75 points (range 6.5–7) (Supplementary Table S3). Points were lacking in quality criteria C7 reporting if a clear accountability of the confounders implied in multivariate analyses. Two studies did not registered whether they performed multivariate evaluation to address confounder. A positive correlation (R ​= ​0.8454) was reported comparing study size and the impact factor of the journal (p ​= ​0.033) (Supplementary Fig. S1(A)). There was no correlation (R ​= ​0.058) among the study quality evaluated by the REMARK criteria and impact factor (p ​= ​0.601) (Supplementary Fig. S1(B)).

We assessed RoB in included studies using QUIPS and found that one study is rated as ‘High Risk of Bias’. All studies were deemed eligible in the review notwithstanding of their score and their limitations were recorded. The quality appraisal score for each research paper is projected in Supplementary Table S4.

To check publication bias among the included studies, we planned to construct funnel plot and perform egger’s test (Fig. 3A and B). however, number of eligible studies were less 10. Hence, only funnel plot is presented but egger’s test was carried out. The quality of the evidence (GRADE) was ‘Moderate’ as it was downgraded for inconstancies and imprecision. The overall quality of the studies analyzed OS and DFS was ‘Moderate’ (Supplementary Table S5).

Fig. 3.

Funnel plots of studies included in the 3 meta-analyses: A) overall survival, B) disease-free survival.

4. Discussion

Improvement in OSCC survival rate is paramount important due its high incidence of morbidity and mortality rate. Over last 5-6 decades, from surgery to immunotherapy with variable success is suggested to improve it. Nevertheless, even after advancement of surgery field, overall survival period in OSCC is limited.¹⁰ Besides that, other clinico-pathological factors such as age, gender, location, tumor grading, and histopathological features/grades have influence over survival rate.¹¹ Survival rate based on above factors range within 40%–60% and goes lower cumulatively with addition of these factors. Several biomarkers were studied to find out reliable prognostic or predictor factors for better survival and disease free outcome in OSCC. Recently, several authors studied TATE to predict survival in OSCC.¹² However, OSCC studies, similar to breast, melanoma, pancreatic, colon-rectal malignancy has found contrasting association between eosinophil counts in tumor tissue/TATE and survival prognosis.^12,13

This systemic review and meta-analysis envisioned that an raised level of TATE could forecast better survival in patients with OSCC. A high count of TATE was significantly related with increased overall survival in pooled meta-analysis. Elevated level of TATE could be a strong prognostic factor for OS. Eosinophils are part of host immune response during tumorigenesis and elevated levels of eosinophils are indicative of immune response in early stages. Eosinophils produces growth factors, vascular permeability factor and granulocyte macrophage colony stimulating factor, cytokines (IL-8, IFNγ and TNF-α, IL-4, IL-5 and IL-13), eosinophil granule proteins, major basic protein, eosinophil derived neurotoxin, eosinophil peroxidase. Eosinophils are recruited to tumor microenvironment (TME) through complex set of chemicals and under the influence of other immune complex cells.¹⁴ Various mechanisms have been explained specially IL-5 and eotaxin levels although other pathways can recruit them. They perform the tumericidal activity through different mechanism in TME.¹⁵ Elevated eosinophil cationic protein (ECP) bind to cell surface and destabilize the cell membrane by introduction of pores, there by altering the osmotic pressure resulting in death of tumor cell.¹⁶

Despite eosinophils have these beneficial effects and increases the survival, other studies have reported a poor survival outcome in OSCC. The factors responsible for variable outcome in high TATE cases could be different (clinical/pathological) stages of OSSC,¹⁷ histological differentiation or grades, absence/presence of metastasis.¹⁸ There is increase in eosinophil counts from stage I or early to stage IV or advanced stage.¹⁹ Oliveira et al.²⁰ reported that early stage OSCC patients with early stages (T1/T2) presented absent/mild TATE. In contrary intense eosinophilia was strongly related with tumors in the advanced stages (T3/T4). However, same study reported no correlation with node positivity. As we know that, in advance stages, tumor shows loco-regional spread to adjacent muscles and distant spread to lymph nodes thereby reduces the survival. Rakesh et al.²¹ studied invasive tumor front area (invasive) and reported better survival outcome. Tumor front is an area where tumor invasion is active hence; it can be said that, since body immune cell response will be higher at these location, survival should be better and similar to colorectal carcinomas.¹² However, higher levels of eosinophils at invasive front have been linked to stromal invasion and poor prognosis. These studies have not elaborated on possible mechanism or explanation and thus require investigation.

The lymph node metastasis or lymph node status is an important factor for predicting the survival in OSCC. Studies have reported week association between mean eosinophils count in same patient primary tumor and metastatic lymph node.²² Similar results were also seen in Aghbali et al.²³ and Oliveira et al.²⁰ However, Oliveira et al.,²⁴ in their study involving early OSCC (T1/T2) found strong association between TATE with lymph node status suggesting the higher risk of neck recurrence/metastasis in OSCC. Jain et al.¹⁸ observed that OSCC without metastasis have higher mean eosinophilic counts as compared to OSCC with metastasis. Probably, spread to lymph node or presence of eosinophils could involve another still unknown mechanism.

In addition to the above factors, it was observed that histological grade or differentiation have correlation with overall survival.²⁵ Furthermore, certain correlation between squamous differentiation and eosinophilic counts were reported in the literature.^17,18,25,26 In addition to that, higher eosinophilic count in well-differentiated OSCC than moderate differentiated OSCC was observed.¹⁸ Iwasaki et al.,²⁷ however, had different observation about histological differentiation with low degree (poorly differentiated) of tumor cell differentiation having strong eosinophilic infiltration. Peurala et al.²⁸ reported higher mean eosinophil counts in moderate and poorly differentiated OSCC as compared to well differentiated type. But the relationship between metastasis and histological differentiation with eosinophilic immune response is still elusive.

In addition to overall survival, our study demonstrated that pooled meta-analysis of DFS has poor correlation with high TATE in tumor tissue. Early evidence on recurrence showed that it is responsible for increase mortality, significantly reduces the survival and worsen the prognosis.²⁹ Majority OSCC cases of high TATE in Rakesh et al. study developed loco-regional recurrence. Dorta et al.,²⁶ however, found that recurrence is less in patients with high TATE count. Difference between studies could be attributed to tumor tissue location. Higher eosinophil levels at tumor front have been implicated and observed to play role in tumor loco-regional invasion.³⁰ However, the present systematic review observed the association of high TATE with increased risk of recurrence in OSCC patient. This aspect of the study result could be explained by the fact that eosinophil have dual functionality in tumor tissue.

In the present review, majority of studies reported patients with OSCC at different sites in the oral cavity and majorly lacks the prognostic data. Moreover, there is heterogeneity in terms of the tumor compartment analyzed for the investigation of TATE. (Table 1). However, the reporting of these compartments was mostly ambiguous, making it impossible to compare the results. We also observed variations in terms of the type of surgical samples (incisional vs excisional) used for the TATE analysis.

Although our study results showed favorable association with high TATE counts, these results need to seen carefully as many studies, especially those performed in OSCC were carried out in different setting (without case control studies design; no age-match, no sex-match). Studies carried out on TATE estimation in tumor tissue using routinely practiced light microscopy and HE stain concluded that there are no extra costs for staining or pathological processing. Hence, it can be performed as part of routine histopathology reporting. It is important to note that in spite of positive outcome reporting from studies, there was variability in TATE categorizations and counting methodology. Furthermore, studies counted eosinophils either ‘mm^2’ or ‘per high power field’. Clearly, there is need for standardization through more prospective clinical studies with definitive criteria. Although results in mm square gives more accurate results, we recommend that outcomes should be measured with ‘per high power field’ as morphometric software may not be available with every pathology practice.

5. Conclusion

In conclusion, present study attempted to find role of TATE as predictor of overall survival and disease free period in OSCC. High tissue eosinophilia showed positive overall survival outcome and negative disease free survival in OSCC. TATE could be introduced as part of histology reporting after thorough standardization and validated with clinical trials with large cohort. TATE is easily noticeable on routinely performed histotechnique and hence can be performed in day today practice. However, standardization in counting and reporting methodology is warranted in future studies.

Funding source

None declared.

Declaration of competing interest

None declared.

Appendix A. Supplementary data

The following is the Supplementary data to this article: