Hao-Ran Zhang,Xue Yang,Cui-Hong Zhu,Dan Chen,Xiong-Zhi Wu

1Tianjin Hospital of Integrated Traditional Chinese and Western Medicine,Tianjin 300100,China;2Department of Medical Oncology,Tianjin Medical University General Hospital,Tianjin 300052,China;3Tianjin Cancer Hospital Airport Hospital,Tianjin 300308,China;4Tianjin Jingui Linglan Science and Technology Ltd.Tianjin 300000,China;5Tianjin Medical University,Tianjin 300384,China.

Abstract

Key words:Colorectal cancer,Tongue coating thickness,Prognosis,EGF,lactic dehydrogenase,Diagnostic markers

Background

According to World Health Organization,colorectal cancer (CRC) is the third most common cancer in the world [1].Although join point trend analysis shows that the incidence rate for CRC declined with a relatively large contribution from screening that can detect and remove precancerous polyps [2,3],nearly 25% of patients with stage Ⅱ may relapse or develop metastases.Though chemotherapy,targeted therapy,radiofrequency ablation and other new progressed therapeutic methods are applied in the treatment of pulmonary metastases from CRC,and the 5-year overall survival rate is still no more than 5% by systemic therapy [4].Thus,in order to reduce mortality of CRC,early detection is most crucial.Recently years,serum tumor markers such as CEA and CA19-9 are well-known tumor markers that are commonly used in restaging and surveillance of CRC clinics.However,they are not specific for early detection of CRC [5,6].The prognosis assessment and condition judgment of the CRC remains a challenge clinically.Therefore,identification of diagnostic markers to evaluate the prognosis of CRC clinically should be urgently developed.

Traditional Chinese Medicine (TCM) suggests that the tongue coating thickness (TCT) is a very sensitive index that reflects the progression and degree of diseases.The main aim of this study is to observe whether TCT can be an indicator to predict prognosis of CRC.A conventional method to evaluate TCT involves direct examination of a protruded tongue by a practitioner.However,the result is not only affected by experience,knowledge,and diagnostic skills of individual practitioners but also by environmental factors such as light,brightness [7–9].To overcome the above limitations,we propose a quantitative standard for judgment on thin coating and thick coating by computerized imaging technology and constructing a regression equation.

Materials and methods

Ethical statement

This study was approved by the ethics committee of Tianjin Medical University Cancer Institute and Hospital,National Clinical Center for Cancer,Tianjin,China (No.bc2015003).Informed consents were obtained from all participants involved in this study.

The composition and working procedures of this Ethics Committee are in line with the International Conference on Harmonisation Good Clinical Practice and China’s Good Clinical Practice Principles and complied with the relevant national laws and regulations.The exam was in line with China’s Health Commission,Involving People’s Biomedical Research Ethical Review Approach (2016),Guidelines for the Quality Management of Drugs for Clinical Trials (2010),the National Food and Drug Administration,the Guidelines for Ethical Review of Drug Clinical Trials (2003),the Medical Device Clinical Trials Regulations (2004),the WMA Declaration of Helsinki,the International Council on Medical Science Organizational Committee International Ethical Guidelines for Biomedical Research Involving Human Subjects.

Patients

Seventy-four patients with CRC were collected through the outpatients of Tianjin Medical University Cancer Institute and Hospital from May 2010 to September 2011,in which there were 49 patients confirmed with recurrence or metastasis.Pathological testing for all patients was performed to confirm CR.Contrast enhanced CT,MRI,PET-CT or biopsy were performed to confirm their local recurrence or distant metastasis.All photos of patients’ tongue were taken with a SONY camera in the same room and under constant conditions such as brightness or distance.Patients were informed to avoid mouth rinsing as well as food and liquid intake for 4 h.

Inclusion criteria of sample set.1.Diagnosed as CRC by a pathological test both males and females were included.2.Colonic recurrences were usually diagnosed by colonoscopy.Contrast enhanced CT,MRI,PET-CT or biopsy was performed to detect distant metastases.3.Patients were informed that the detail information of face has not been collected on photos.

Tongue photos of training set.To develop the risk score model,40 tongue images that is literally the absence of tongue coating and 40 tongue image images that tongue body were not visible at all were selected as the training set.

Inclusion criteria of training set.1.Diagnosed as CRC by a Pathological test and both males and females were included.2.Patients were informed that the detail information of face had not been collected on photos.3.Thick tongue coating or without tongue coating tongue images were selected in the training set.

Quantitative indices

According to the principle of colorimetry,any color is composed of three primary colors by mixing an appropriate proportionany.Three primary colors of light,R (red),G (green) and B (blue),make up the color image.The measurement for color of object can be simplified as the measurement for strength of the degree and the ratio between R,G,B three primary colors of object.Tongue color is also consisted by RGB colors.We used Adobe Photoshop CS5 to analyze five points on patients’ tongues randomly;and collected the data of the RGB values.All tongues images were analyzed by Photoshop.

Immunohistochemistry

Tongue tissue was derived from 4-month-old male SD rats.Formalin-fixed,paraffin-embedded tissue specimens were cut into 4-μm sections.The slides were deparaffinized in xylem and dehydrated in a graded ethanol series,and the sections underwent antigen retrieval in citrate solution for 2 min and 30 s.Endogenous peroxides was blocked with 3% hydrogen peroxide for 10 min,then the sections we rewashed with phosphate-buffered saline.After blocking,they were incubated overnight with primary antibodies (1:400) (Anti-EGFR antibody,abcam,ab231) diluted by 1% BSA,and then second antibodies for 30 min.Antigen staining was performed using DAB horseradish peroxides color development kit and then counterstained with hematoxylin.The immunoreactivity of proteins in each tissue core was assessed independently by two experienced pathologists.The intensity of staining was scored from 0 to 3 (0:absent,1:weak,2:intermediate,3:strong staining) and the extent of staining was scored from 0 to 100%.The final quantization of each staining was obtained by multiplying the 2 scores.

Statistical analysis

Regression equation predicting thickness of tongue coating was constructed using binary logistic regression analysis.The optimal cut off of probabilities to diagnosis thick tongue coating was determined by receiver operating curve analysis.χ2test for paired data and kappa test were used to determine the diagnostic value for recurrence and/or metastasis in CR patients.Kolmogorov-Smirnov test was used to determine the distribution of ALT,AST,ALP ALB,TP,GLO,TBIL,DBIL,GGT,LDH,GLU,UA,CA724，CA199，CA242 and CEA.Data with the skewed distribution were presented as median (quartile interval).The association between the thickness of tongue coating and clinico-pathological character was evaluated by chi square test and twoindependent-sample test.The two-independentsamples χ2was used to determine whether there were significant differences in the thin coating and thick coating between patients with recurrence and/or metastasis and patients without recurrence and/or metastasis.Kaplan-Meier method was used to analyze survival time.Statistical analysis was performed by SPSS (version 16.0).All data were expressed as mean ± standard deviation.

The Minimum Standards of Reporting Checklist contains details of the experimental design,and statistics,and resources used in this study.

Results

Quantitative analysis of detected thickness of coating

The tongue features of the subjects participating in the study were represent with the RGB color space.Training set was composed of no coating and thick coating which are two distinct groups (Figure 1),thus,logistic regression can be used to predict the outcome of a categorical dependent variable based on RGB [10].Logistic regression by utilizing these 3 tongue features is significant differences.We removed one of the 3 tongue features which did not show the most significant differences (P>0.05) and performed logistic regression two times.Among them,we can obtain a predicting equation for the probability of belonging to no coating or thick coating.An accuracy of 78% was reached through the logistic regression model (LRM).

P= 1/(1 + e?(6.948?0.977r+1.526g?0.276g))

The tongue features of sample set were predicted using the model.We calculated an optimal cut-off level for the probabilities to diagnosis thin coating or thick coating by receiver operating curve analysis (Figure 2).To evaluate the standard of classification,3 experienced clinicians were involved in the present study.They had been instructed in the conventional method to label each image as either “No coating”,“Thin coating”,or “Thick coating” (Figure 1).If there was a disagreement,we chose the majority to establish the final diagnosis [10].The percentage of tongue coating was measured by the LRM and TCT was classified as either thin coating,or thick coating according to the existing diagnostic criteria (Table 1) [12].Finally,we evaluate the diagnostic agreement between the results of the LRM and TCT.The result of LRM was consistent with visual recognition,kappa was 0.835.

Characterization of patient with and without thick coating

Thirty-six males and 38 females were included in the study.The age of all patients ranged 26–76 years with a median age of 58 years.No significant difference was observed in age,gender,histotype,primary tumor site,primary tumor stage,lymph node metastasis and TNM between patients with thin coating and patients with thick coating (P>0.05).

Diagnostic value of the thickness of tongue coating for prognosis of CRC

Patients with thick coating had a higher risk of recurrence or/and metastasis than those with thin coating (χ2= 13.371,P<0.001).The χ2test for paired data proved that there was no significant difference in diagnosis of recurrence or/and metastasis between the thickness of tongue coating and the traditional diagnostic criteria (P＞0.05).Kappa was 0.417 (P<0.001).The sensitivity,specificity,positive predictive value and negative predictive value were 63.39%,76.00%,85.00% and 55.88%,respectively (Figure 2).The median overall survival (OS) of patients with thin tongue coating was longer than the patients with thick tongue coating (79.3 and 57.9 months,respectively,P= 0.06 in Figure 3).

Diagnostic value of serological examination for thickness of tongue coating within one month before or after their photos taken.

No significant difference was found in serum tumor markers (CA724,CA199,CA242,CEA)，and liver function index (ALT,AST,ALP ALB,TP,GLO,TBIL,DBIL,GGT,GLU,UA) between patients with thin coating and patients with thick tongue (Table 2).LDH level was higher in patients with thick tongue coating than the patients with thin tongue coating (P= 0.01).The thickening of tongue coating of patients with colon cancer was significantly associated with chemotherapy (P= 0.001).LDH level is associated with survival in patients with CRC and may be used as an indicator of future survival of cancer patients.

Immumohistochemical staining in five different parts of the tongue of the rat

We divided the rat’s tongue into five regions,the tip,middle part,left side,right side and root of the tongue (Figure 4A).The expression of epidermal growth factor (EGF) receptors (EGFR) in these five regions was detected.The results showed that the expression of EGFR receptors in the root of the tongue was the most (Figure 4B),so the formation of thick tongue greasy coating was likely to be closely related with the level of serum EGF in vivo.The results imply that the serum levels of EGF in patients with the thick coating might be elevated.Therefore,the survival time of CRC patients with thick coating was shorter,it is likely that EGF could simultaneously promote the growth of tongue coating and colorectal tumor progression.This study was approved by the ethics committee of Tianjin Medical University Cancer Institute and Hospital,National Clinical Center for Cancer,Tianjin,China (No.20162).

Discussion

Tongue diagnosis is the major diagnostic method in TCM that was developed thousands of years ago [11].Tongue diagnosis contains examination of the tongue body (its color,shape,moisture,and movement) and coating (its color,thickness,and distribution).Tongue coating thickness is considered to reflect the progression and degree of diseases [12].A conventional method to evaluate TCT involves direct examination of a protruded tongue by a practitioner.However,the result is not only affected by his or her subjectivity but also by environmental factors such as experience,knowledge,and diagnostic skills of individual practitioners [12,13].To overcome the above limitations,different types of tongue diagnosis systems have been recently developed [12].In this study,we propose a quantitative standard for judgment on thin coating and thick coating by computerized imaging technology and constructing a regression equation.The proportional odds model was used for developing the diagnostic standard of the TCT.Eighty-six-point two nine percent was the cutoff point to divide into thin coating and thick coating.The degree of correlation between the TCT judgments and the gold standard established by assessors was 0.835.

Table1 Variables in the regression equation

Table2 Serological examination of the patients with CRC

According to TCM theory,the tongue coating is produced by Wei Qithrough fumigation [10].Therefore,the alteration of the tongue coating can be closely related to the progress of gastrointestinal diseases.Dramatically,our data shown patients with thick coating had a higher risk of recurrence or/and metastasis than those with thin coating.The sensitivity,specificity,positive predictive value and negative predictive value were 63.39%,76.00%,85.00% and 55.88%,respectively.Moreover,the median OS of patients with thin tongue coating was longer than the patients with thick tongue coating (79.3 and 57.9 months,respectively).Thus,the thickness of tongue coating could be used as an indicator for monitoring the recurrence of CRC after radical operation.

According to TCM theory,the tongue coating is produced by Wei Qi (similar to the function of digestive system in Western medicine) through fumigation [10].The tongue coating changed from thin to thick,indicating the disease has aggravated from superficies to the interior [15].It was considered that the alteration of the tongue was closely related to diseases of the digestive system,especially intestinal diseases [16].

Tongue mucosa consisted mainly stratified squamous epithelium cells which include keratinocyte layer,granular layer,prickle cell layer,basal cells.Bacteria,food residues,and desquamated keratinized epithelium which originated chiefly from the filiform papilla were the main material of tongue coating formation [17–19].The dorsal surface of the mammalian tongue is covered with 4 kinds of papillae,fungiform,circumvallate,foliate and filiform papillae.Modern research has shown that reproductive activity enhancement or educed apoptosis of glossal epithelial cells was the main characteristics of thick greasy tongue fur formation.

Many molecules,including growth factors,transcription factors,extracellular matrices,cell surface receptors,and matrix degrading enzymes,have been found to be involved in this process.EGF is involved in epidermal regeneration by stimulating the proliferation and migration of keratinocytes at the wound edge through its interaction with high affinity receptors on both fibroblasts and keratinocytes [20].There is growing evidence that EGF regulates a variety of biological functions.Through the autocrine or paracrine mechanism in the epidermal growth factor receptor EGF-R,EGF lead to filiform papillae over-proliferation and the reduction of the apoptotic of tongue mucosal epithelial cells,which finally resulting in the formation of pathological thick coating.

What’s more,EGFR has been shown to be overexpressed in colon cancer cell lines and is detectable by immunohistochemistry in 65–75% of CRC tumors [21].EGFR expression progressively increases with malignant transformation from normal colon,through adenoma,to the poorly differentiated and metastatic cancer.Over-expression of EGFR is associated with poor prognosis of cancer in the majority of studies [22–25].Thus,the thick tongue coating may imply the increase of serum EGF levels in the body.

Conclusion

In conclusion,the current research provides preliminary evidence that thick tongue coating was related to poor prognosis of tumor,at least CRC.Patients with thick coating had a higher risk of recurrence or/and metastasis and shorter survival time than those with thin coating.It also elucidates possible mechanisms of formation of thick coating on the root of the tongue:EGF could possibly lead to the thickening of tongue coating.Therefore,our investigation provided new opinions for further studies about the relationship between tongue coating changes and tumor progression.