DOI: https://doi.org/10.21203/rs.3.rs-1925097/v1
Background: With the depth research of tumor microenvironment, tumor stroma was considered to play a leading role in the tumor malignant behavior, the PD-L1 was also related with the tumor stroma. Tumor–stroma ratio (TSR) has already been regarded as a novel prognostic factor in many cancers. Our study aims to assess the TSR and PD-L1 clinical value in the HCC patients.
Material and methods: 95 patients who diagnosed HCC, were included in our res. TSR was estimated on HCC specimen HE sections, and the optimal TSR cut-off value was determined by ROC curves. The correlation between TSR and clinicopathologic features was also culculated. Immunohistochemistry staining was also carried out to analysis the PD-L1 expression levlel in HCCs.
Results: The optimal TSR cut-off value was 0.525. The median OS of the stroma-high and stroma-low groups was 27 vs. 36 momths. The median RFS of the stroma-high and stroma-low groups was 14.5 vs. 27 months. In Cox multivariate, TSR was an independent prognostic factor in HCCs overall survival (OS) and recurrence free survival (RFS) who underwent liver resection. Immunohistochemistry (IHC) staining revealed that TSR-high HCC samples with high PD-L1-positive cells expression.
Conclusions: Our results suggest that TSR can predict the prognosis in the HCCs who underwent liver resection, the TSR has relation with the PD-L1 expression and may be the therapy target which can dramatically improve the HCC patients’ clinical outcomes.
Cancer cells and tumor microenvironmrnt are have significant tumor heterogeneity, which paly a decisive role in the poor prognosis of cancer patients. The tumor stroma is the main component of the tumor microenvironment (TME), which provides supportive and permissive conditions for tumor invasion and metastasis. TSR can predict the expression of PD-L1 in HCCs and may a biomarker of PD-L1 immunotherapy.
Hepatocellular carcinoma (HCC) is the most malignant tumor in the digestive system and has become the third most common cause of cancer-related death worldwide [1]. The burden of HCC is increasing globally; there could be 1 million patients with liver cancer by 2030 [2], and the number of patients with HCC patients in China has been increasing mainly due to hepatitis virus infection. the treatment of HCC includes hepatectomy [3, 4], liver transplantation[5], radiofrequency ablation[6], transcatheter hepatic arterial chemoembolization (TACE), targeted drugs, and immune therapy[7], among others. Hepatectomy and transplantation are regarded as curative treatments for this disease [3]. Strategies have been made to improve HCC treatments and diagnosis, but the clinical outcomes of patients with HCC remains poor, with high recurrence rates [8] and high mortality. Thus, the accuracy of HCC prediction, especially for patients who undergo hepatectomy, must be further improved to obtain a better treatment effect.
Advanced research has shown that stroma cells in cancer play a major role as an important modulator of tumor cell growth, pathogenesis, and progression; these cells also have the potential to influence prognosis in patients with cancer [9, 10]. The tumor stroma is an important form of tumor microenvironment (TME), which provides supportive and permissive conditions for tumor invasion and metastasis [11, 12]. Additionally, interactions between tumor cells and stromal cells result in the production of different cytokines and enzymes that play important roles in tumor growth and progression.
Tumor heterogeneity has a decisive role in the poor prognosis of cancer, especially in HCC, which has significant heterogeneity [13, 14]. Tumor heterogeneity is associated with TME. TME also participates in epithelial-mesenchymal transition (EMT), which is an active modulator of tumor invasion and metastasis and leads to worse clinical outcomes. Moreover, advancements in research have determined that the tumor stromal can intervene in EMT signaling. In addition, increased is tumor stroma associated with transforming growth factor-β (TGF-β) expression. Previous studies indicate that itself is mutated and subsequently resulted in PD-L1 aberrant expression [15]. Peritumoral stroma inactivation of PD-L1 affects the HCC poor prognosis and determinant of resistance to immunotherapy [16]. Therefore, the observation of PD-L1 expression in tumor stromal is highly unusual. However, the relationship of PD-L1 and TSR in human HCC remains unknown.
The purpose of our research is to analysis the prognostic value of the tumor–stroma ratio (TSR) grading and PD-L1 expression in HCC patients who underwent hepatectomy and to explore its relationship with other prognostic factors.
2.1 Patient selection and data collection
Ninety-five patients who underwent hepatectomy were included in this study from Xiangya Hospital, Central South University, China. No patients in our study underwent TACE, RFA or other nonsurgical treatments. Clinical raw data were collected. All patients signed informed consent, in the process of study, we were cautious to ptotect the patient’s privacy. This study was approved by the ethics committee of XiangYa Hospital, Central South University.
2.2 TSR ratio and score
To obtain an accurate percentage of TSR, we first obtained cancer percentages and then calculated the TSR using the following formula: the cancer percentage plus the tumor stroma was 100%. For example, if the cancer percentage was 60, the TSR would be 40% [17]. We analyzed 5-mm hematoxylin–eosin-stained sections in all tumor samples. To obtain an accurate TSR and scoring for each patient, two researchers assessed the TSR value on all tumor slides in in a blinded manner. Using the 4× objective, we had selected the most invasive part on each HE slide. The researchers evaluated the score of TSR by using a 10× objective. The percentage of tumor mainly ranged from 20 to 80%.
2.3 Immunohistochemistry (IHC)
IHC staining of PD-L1 was performed as described previously [7]. An anti-PD-L1 polyclonal antibody (ABCAM, Cat# ab205921) diluted at 1:500 was used as the primary antibody. The IHC staining were analyzed and scored following full-slide digitalization with the Pannoramic Scan and the database-linked TMA Modul software (3DHISTECH, Budapest, Hungary). The numbers of PD-L1-positive were counted respectively in seven respective visual fields from tissue samples with high-power fields.
2.4 Follow-up
All HCCCpatients in our study had considerable clinical survival data, including recurrence time and survival data. The end point of follow-up was HCCs death or December 2017. OS time is taken as the time interval from the date of the operation to death. RFS was calculated from the first operation to HCC recurrence.
2.5 Statistical analysis
Chi-square test and t student test were used to assess the differences in patient characteristics. Survival analysis were drawn using Kaplan–Meier methods. Cox regression were used to perform univariate and multivariate analyses in HCC patients, and hazard ratios (HR) with 95% confidence interval (CI) was shown in the part of results[18]. Statistical analyses were performed using Prism software (GraphPad Prism Software, La Jolla, CA) and SPSS 21.0 (SPSS Company, Chicago, IL) for Windows.
To calculate the ideal cut-off value of the TSR, we calculated the ROC curve area by OS and the ROC curve area was 0.68 (95% CI, 0.57 to 0.79), the sensitivity was 34% and the specificity was 87%. The optimal cut-off value was 0.525 (Fig. 1-A). For subsequent experimental analysis, all patients were divided into the following groups: the stroma-high group (TSR > 0.525) and stroma-low group (TSR ≤ 0.525) (Fig. <link rid="fig1">1</link>-B and 1-C). In the stroma-high group, micrometastatic nodules were more frequent (Fig. 2.). We cosidered that the stroma plays a leading role in HCC progression.
In this study, 95 patients were enrolled in our study, and we analyzed the clinicopathological characteristics of these patients in Table 1. The stroma-high group included 26 patients, and the stroma-low group included 69 patients. The ages of the stroma-high and stroma-low groups were (54.85 ± 9.05) years and (50.16 ± 12.31) years, respectively. The tumor sizes of the stroma-high and stroma-low groups were (5.28 ± 3.0) cm and (8.96 ± 3.69) cm, respectively. The platelet counts of the stroma-high and stroma-low groups were (167.15 ± 79.14) 109/L and (173.97 ± 75.68) 109/L. A total of 83 (86.32%) patients had CTP A stage, and 73 (74.74%) patients had single tumors. A high preoperative AFP was observed in 65 (68.42%) patients.
| Clinical character | Stroma high (n = 26) | Stroma low (n = 69) | P-value | ||
|---|---|---|---|---|---|
| Age, years | 54.85 ± 9.05 | 50.16 ± 12.31 | 0.08 | ||
| Serum albumin, g/L | 43.01 ± 6.04 | 441.15 ± 5.36 | 0.25 | ||
| Tumor size, cm | 5.28 ± 3.00 | 8.96 ± 3.69 | 0.00 | ||
| Platelet, 10^9/L | 167.15 ± 79.14 | 173.97 ± 75.68 | 0.00 | ||
| ALT, U/L | 48.18 ± 21.25 | 42.46 ± 28.96 | 0.36 | ||
| AST, U/L | 54.52 ± 20.00 | 48.58 ± 39.57 | 0.47 | ||
| PT, s | 13.06 ± 1.39 | 13.38 ± 1.16 | 0.32 | ||
| Gender | Male | 21 | 59 | 0.39 | |
| Female | 5 | 10 | |||
| HBsAg | Negative | 2 | 12 | 0.23 | |
| Positive | 24 | 57 | |||
| AFP, ng/mL | ≤ 20 | 6 | 24 | 0.27 | |
| > 20 | 20 | 45 | |||
| CTP | A | 19 | 63 | 0.02 | |
| B | 7 | 6 | |||
| Liver cirrhosis | No | 8 | 22 | 0.91 | |
| Yes | 18 | 47 | |||
| Tumor encapsulation | No | 17 | 50 | 0.46 | |
| Yes | 9 | 19 | |||
| Tumor number | Single | 13 | 58 | 0.00 | |
| Multiple | 13 | 11 | |||
| Satellite nodules | No | 22 | 63 | 0.34 | |
| Yes | 4 | 6 | |||
| Edmondson grade | I–II | 16 | 53 | 0.14 | |
| III–IV | 10 | 16 | |||
| BCLC stage | 0 | 0 | 6 | 0.00 | |
| A | 12 | 51 | |||
| B | 9 | 10 | |||
| C | 5 | 2 | |||
| TNM stage | I | 6 | 48 | 0.00 | |
| II | 11 | 10 | |||
| III | 9 | 11 | |||
| HBsAg, hepatitis B surface antigen; AFP, α-fetoprotein; TNM, tumor-node-metastasis; PT, Prothrombin time; CTP, Child-Turcotte-Pugh; BCLC stage: The Barcelona Clinic Liver Cancer staging; ALT, glutamic-pyruvic transaminase; AST, glutamic oxalacetic transaminase. | |||||
TSR levels were closely correlated with tumor size, CTP stage, TNM stage and BCLC stage (P < 0.05). No obvious correlations with gender, HBsAg, liver cirrhosis, serum albumin, glutamic-pyruvic transaminase (ALT), glutamic oxaloacetic transaminase (AST), etc. (P༞0.05) were found.
Kaplan–Meier method were used to analysis the association between the TSR and RFS/OS (P < 0.01; Fig. 3). The median OS of the stroma-high and stroma-low groups was 27 momths (95% CI, 0.45 to 1.25) and 36 months (0.80 to 2.23), respectively. The median RFS of the stroma-high and stroma-low groups was 14.5 months (95% CI, 0.33 to 0.88) and 27 months (95% CI, 1.14 to 3.05), respectively.
In the univariate analysis between OS and all clinicopathologic characteristics, we found that TSR (HR, 3.33, 95% CI, 0.093–0.97; P = 0.04), tumor size (HR, 1.33, 95% CI, 1.13–1.58; P < 0.001), and BCLC stage (HR,4.93, 95% CI, 1.35–18.02; P = 0.02) were significant indicators for the OS (Table 2). And in the multivariate analysis, the TSR (HR, 8.34, 95% CI, 1.18–59.48; P = 0.03) was an independent prognostic factor. In Cox multivariate analysis of RFS, the TSR was also an independent prognostic factor. The HR of the TSR was 10.06 (95% CI, 1.61–63.03; P = 0.02) (Table 2).
| Clinicopathologic variable | RFS | OS | ||||
|---|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |||
| Univariate analysis | ||||||
| Gender (male vs. female) | 1.38(0.28–6.81) | 0.69 | 2.04(0.53–7.84) | 0.30 | ||
| Age, years (> 60 vs. ≤60) | 1.02(0.98–1.07) | 0.32 | 1.02 (0.98–1.06) | 0.40 | ||
| Serum albumin, g/L (≤ 35 vs. >35) | 1.03 (0.93–1.14) | 0.61 | 1.02(0.94–1.10) | 0.68 | ||
| Platelet,10^9/L (≤ 160 vs. >160) | 0.997(0.99–1.004) | 0.12 | 1.001(0.996–1.01) | 0.63 | ||
| ALT, U/L (≤ 50 vs. >50) | 0.98(0.981–1.007) | 0.35 | 0.994(0.978–1.01) | 0.44 | ||
| AST, U/L (≤ 40 vs. >40) | 0.994(0.99–1.03) | 0.29 | 1.004(0.99–1.017) | 0.59 | ||
| PT, s (≤ 13.2 vs. >13.2) | 0.64(0.40–1.02) | 0.06 | 0.82(0.57–1.17) | 0.27 | ||
| AFP, ng/mL (> 20 vs. ≤20) | 3.84(0.81–18.14) | 0.09 | 1.11(0.44–2.85) | 0.82 | ||
| HBV (presence vs. absence) | 2.96(0.36–24.37) | 0.31 | 1.78(0.56–5.69) | 0.33 | ||
| TSR (> high vs. ≤low) | 1.16(0.34–4.06) | 0.82 | 3.33(1.03–10.73) | 0.04 | ||
| BCLC stage (C vs. 0/A/B) | 1.78(0.46–6.84) | 0.40 | 4.93(1.35–18.02) | 0.02 | ||
| TNM stage (II/III vs. I) | 4.75(0.59–38.36) | 0.14 | 3.19(0.86–11.87) | 0.08 | ||
| Tumor number (multiple vs. single) | 1.44(0.44–4.65) | 0.55 | 1.53(0.54–4.36) | 0.42 | ||
| Edmondson grade (III/IV vs. I/II) | 1.16(0.34–3.98) | 0.82 | 1.21(0.47–3.15) | 0.70 | ||
| Tumor size, cm (> 5 vs. ≤5) | 1.13(0.95–1.34) | 0.17 | 1.33(1.13–1.58) | 0.01 | ||
| Satellite nodules (presence vs. absence) | 1.27(0.24–6.62) | 0.78 | 4.66(0.56–38.6) | 0.15 | ||
| Tumor encapsulation (none vs. complete) | 1.56(0.51–4.79) | 0.44 | 1.04(0.40–2.67) | 0.94 | ||
| Liver cirrhosis (presence vs. absence) | 3.84(0.81–18.14) | 0.09 | 1.41(0.54–3.67) | 0.49 | ||
| CTP (A vs. B) | 2.69(0.32–22.28) | 0.36 | 6.57(0.81–53.07) | 0.08 | ||
| Multivariate analysis | ||||||
| PT, s (≤ 13.2 vs. >13.2) | 0.91(0.76–1.10) | 0.34 | NA | |||
| CTP (A vs. B) | NA | 1.11(0.52–2.37) | 0.79 | |||
| AFP, ng/mL (> 20 vs. ≤20) | 1.29(0.79–2.09) | 0.31 | 1.13(0.67–1.94) | 0.63 | ||
| Tumor size, cm (> 5 vs. ≤5) | 1.00(0.92–1.08) | 0.35 | 1.05(0.97–1.14) | 0.23 | ||
| TSR (> high vs. ≤low) | 10.06(1.61–63.03) | 0.02 | 8.34(1.18–59.48) | 0.03 | ||
| BCLC (C vs. 0/A/B) | NA | 2.28(1.08–4.82) | 0.31 | |||
| TNM (II/III vs. I) | 1.89(0.92–3.88) | 0.81 | 1.01(0.44–2.32) | 0.97 | ||
| Liver cirrhosis (presence vs. absence) | 1.12(0.68–1.83) | 0.65 | 1.14(0.66–1.95) | 0.63 | ||
| HBsAg, hepatitis B surface antigen; AFP, α-fetoprotein; TNM, tumor-node-metastasis; PT, Prothrombin time; CTP, Child-Turcotte-Pugh; BCLC stage: The Barcelona Clinic Liver Cancer staging; ALT, glutamic-pyruvic transaminase; AST, glutamic oxalacetic transaminase. | ||||||
In the stroma low group, the PD-L1 high expression rate was 28.9%, and in the stroma high group, the PD-L1 high expression rate was 46.2%, there was significant differences between the stroma-high and stroma-low groups (P = 0.03, Fig-4).
In this study, we clearly found that the TSR might be looked upon as a novel biomarker in the predict factor of patients with HCCs after operation. In the present study, patients with a high TSR had a poor prognosis, and patients with a low TSR had good outcomes. Additionally, the TSR in HCC correlated with invasiveness and metastasis. Therefore, the TSR is a significance prognostic factor for patients with HCC who undergo hepatectomy.
In the past studies, TSR has been looked upon as an independent prognostic factor in many other types of cancers. Wang Kai reported that the TSR was an independent element in esophageal squamous cell carcinoma [19]. Van Pelt GW found that the TSR ratio has an important impact on the biological role and prognosis of colon cancer in patients[20]. Roeke. T has used the TSR percentage to predict breast cancer clinical outcomes [21]. The TSR is also widely used in cervical carcinoma [22] and nasopharyngeal cancer [23].
The tumor stroma, including cancer-associated fibroblasts, immune cells[24], epithelial cells, extracellular matrix (EMC) and extracellular molecules, can promote tumor invasion and metastasis[25]. The interactions between stroma cells and tumor cells activate various molecular signaling pathways, which enable these cells to obtain abnormal phenotypes or functional transformations, as well as change the tumor stroma and promote tumor recurrence and metastasis.
Tumor-associated fibroblasts (CAFs) are an important component of the tumor stroma. CAFs can secrete a variety of cytokines, growth factors, and inflammatory mediators that promote cancer cell proliferation, angiogenesis and EMT, ultimately enhancing the cancer’s ability to invade and metastasize. Subramaniam KS has confirmed that CAFs can promote cancer growth via activation of the interleukin-6/STAT-3/c-Myc pathway [26]. Additionally, Carstens JL found that the stroma can affect prostate cancer cell progression by FGFR1-WNT-TGF-β signaling [27]. According to many researchers, CAFs stimulate the proliferation of tumor cells and participate in tumor angiogenesis through TGF-β [28].
In China, patients with HCC mostly develop the disease from hepatitis infection, and they are often diagnosed with liver fibrosis or even cirrhosis. The severity of cirrhosis is significantly correlated with the survival time of patients with HCC after hepatectomy[18]. However, the patient number in our study was too small to reach this conclusion. We hypothesized that patients with severe cirrhosis may have a high TSR and are more likely to experience metastasis, ultimately leading to a poor prognosis. Therefore, TSR analysis for patients with HCC, especially those with fibrosis or cirrhosis, is more important, and the TSR can provide more accurate guidance for the management of patients with HCC after hepatectomy.
Tumor and stromal cells have mutually beneficial interactions. The growth of the tumor stroma provides necessary support for tumor cells, while stromal cells enhance the malignant biological behavior of tumor cells. We speculated that patients with stroma-rich HCC may be associated with EMT, microvascular invasion, and poor prognosis after liver resection. Stroma cells activate EMT and angiogenesis, which enables tumor cells to enter the blood vessels easily, contributing to vascular invasion and other forms of invasion.
Tumor-associated stromal myofibroblasts are essential for the metastatic progression and immune surveillance escape of solid tumors, including HCC [29]. Moreover, we report findings indicating that TGF-β reduces the expression of the proinflammatory factors CCL4 and interleukin-1β (IL-1β in human ex vivo treated HCC tissues. While this is consistent with the anti-inflammatory properties of TGF-β, whether it is an outright tumor promoter or suppressor is still a matter of some debate. In addition, we describe an inhibitory effect of TGF-β on the secretion of CCL2 and CXCL1 by HCC-derived fibroblasts, which suggests the existence of an indirect stroma-mediated functional link between PD-L1 and immunity [30].
In this study, in patients with a high TSR, micrometastasis nodules were more likely to be detected by microscopy. Thus, we proposed the following hypothesis: (1) the increased TSR of the TME may promote tumor metastasis because the tumor stroma provides more nutrients and growth factors necessary for migration; the stroma cells also prepare the most appropriate “soil” for tumor cells. (2) Compared with a low TSR, a high TSR may enhance malignant biological behaviors in cancer cells, and these tumors may be more prone to metastasis. (3) The high percentage of stroma may provide greater protection for cancer cells from immune cell killing or enable cancer cells to escape and the immune system and aid in therapeutic resistance. (4) PD-L1 expression on tumor- stroma increased with disease progression. This potential mechanism may account for the poor clinical prognosis in patients with HCC who have a high TSR.
In this research, we can confirm that the TSR is an independent factor predicting outcomes in HCC patients who undergo hepatectomy. PD-L1 expression is related to TSR, and the tumor stromal may provide new target for the HCC treatment. Our research has the following shortcomings. First, our study was retrospective, and the sample size was small. Thus, statistical bias may exist. Second, the TSR cannot be accurately obtained, which may lead to variations among studies.
The study suggests that TSR can predict the prognosis in the HCCs who underwent liver resection, the TSR has relation with the PD-L1 expression and may be the therapy target which can improve the HCC patients’ clinical outcomes.
TSR, Tumor–stroma ratio; HBsAg, hepatitis B surface antigen; AFP, α-fetoprotein; TNM, tumor-node-metastasis; PT, Prothrombin time; CTP, Child-Turcotte-Pugh; BCLC stage: The Barcelona Clinic Liver Cancer staging; ALT, glutamic-pyruvic transaminase; AST, glutamic oxalacetic transaminase.
Authorship contribution statement
Dong Wang performed the research and collected the clinical data; Jia Luo collected the clincal data and performed following up, YiMing Tao designed the study, performed the figure, and wrote the draft.
Conflicts of Interest: The authors declare that there are no conflicts of interest.
Ethics approval and consent to participate：This study was approved by the ethics committee of XiangYa Hospital Central South University (No.201709984), and the patients provided informed consent. All protocol was performed in accordance with the relevant guidelines and regulations by the Ethics Committee of of XiangYa Hospital Central South University (No.201709984).
Consent for publication: Not applicable.
Availability of data and materials: The data used or analysed during the current study are available from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no potential competing interests in this work.
Funding: This study was supported by grants from the Natural Science Foundation of Shandong Province (No. ZR202103020004), Clinical Medicine plus X Project of Affiliated Hospital of Qingdao University (202102006), National Natural Science Foundation of China (No.81372630).
Authors' contributions: Dong Wang performed the research and collected the clinical data; Jia Luo collected the clincal data and performed following up, YiMing Tao designed the study, performed the figure, and wrote the draft.
Acknowledgements: We gratefully acknowledge all the authors’ works for this paper and all the patients in our study.
Authors' information：Dong Wang, MD, [email protected], Department of Liver Disease Center, The Affiliated Hospital of Qingdao University, Qingdao, Shandong, China.
Jia Luo, MD, [email protected], Department of hepatobiliary surgery, Hunan Cancer Hospital, Changsha, Hunan, China.
YiMing Tao, MD, [email protected], Department of General Surgery, Xiangya Hospital, Central South University, Changsha, Hunan, China.