The prognostic role of C‐reactive protein in patients with head and neck squamous cell carcinoma: A meta‐analysis

Abstract Background The prognostic role of the C‐reactive protein (CRP) in head and neck squamous cell carcinoma (HNSCC) has not been well investigated. This meta‐analysis aimed to evaluate the prognostic relevance of elevated CRP levels in patients with HNSCC. Methods A relevant literature search was performed in PubMed, Web of Science, and Embase up to September 1, 2020. The pooled odds ratio and hazard ratio (HR) with 95% confidence interval (CI) were applied to evaluate the difference in overall survival (OS), progress‐free survival (PFS), and cancer‐specific survival (CSS) between patients with high CRP and those without. The pooled odds ratio (OR) with 95% CI were used to assess the association between CRP and clinicopathological features. Results A total of 17 studies, including 4449 patients, were included. Pooled results showed that an elevated CRP was associated with worse OS (HR = 1.48, 95% CI: 1.24‐1.77), CSS (HR = 1.85, 95% CI: 1.38‐2.46), and PFS (HR = 1.73, 95% CI: 1.38‐2.17). Male patients, lymph node metastases, and higher tumor stage were related to elevated CRP level (OR = 1.67, 95% CI: 1.34‐2.09; OR = 2.40, 95% CI: 1.44‐3.99; OR = 1.39, 95% CI: 1.12‐1.74). Conclusion Our meta‐analysis demonstrated that an elevated pretreatment of CRP indicates poor prognosis in HNSCC. Therefore, CRP is an indicator of the prognosis of patients with HNSCC and can be recommended for assessing prognoses in clinical work.

recent decades, HNSCC incidence has also increased due to the increasing prevalence of smoking. 3 Given that the vast majority of patients have HNSCC in an advanced stage, multiple treatments are required, including surgery, RT, CT, and treatment with biological or targeted agents. Current standard treatments for locally advanced HNSCC involve surgery and postoperative adjustment CRT. However, postoperative CRT has potential adverse effects, including a high burden of toxicity to and morbidity of patients with HNSCC. In addition, it is costly for patients. Determining which cases will maximally benefit from adjuvant CRT after operation is vital. Therefore, new biomarkers that can be used to assess the prognosis of patients with HNSCC and guide clinical treatments must be discovered. 4 Increasing evidence has proven that inflammatory response can promote the development and progression of carcinomas and affect the survival outcomes in patients with HNSCC. To date, some markers, including CRP, are demonstrated to be prognostic markers in various tumors. Over-expressed CRP involves various tumors, including lung, lymphoma, ovary, and multiple myeloma. [5][6][7][8] CRP is a prototypic acute-phase protein secreted by the liver into the blood. Usually, CRP is <2 mg/L in healthy individuals. 9 In response to the acute inflammation or destruction of tissue cells, CRP plasma concentrations display a rapid and pronounced rise in the blood. 10 The synthesis of CRP is stimulated by numerous factors, including bacteria, antigen-immune responses, trauma, and fungi. CRP is also regulated by some pro-inflammatory cytokines, such as interleukin-6 (IL-6), tumor necrosis factor, and interleukin-1. 11,12 Meanwhile, CRP associated with systemic inflammatory processes has been shown to be a prognostic and predictive marker in various solid tumors, including renal, gastrointestinal, and hepatocellular carcinoma. [13][14][15] In general, the ideal biomarker should be sensitive and specific; it must be relatively simple and noninvasive to F I G U R E 1 Flow  detect tumor biomarkers in the blood of patients with cancer. Measuring CRP as a predictor in clinical practice is easy and cost-effective. However, few studies have explored the role of CRP as a predictor of tumor recurrence and prognosis before the treatment of patients with HNSCC. Thus, this meta-analysis aimed to evaluate the correlation between CRP and survival outcomes in patients with HNSCC.

| Search strategy
PubMed, Web of Science, and Embase were searched for information on CRP and the outcomes of HNSCC up to September 1, 2020. Potentially relevant studies were searched using combinations of terms from the following sets of search terms: ("CRP" or "C-reactive protein"), ("head" or "neck" or "head and neck" or "HNSCC" or "nasopharyngeal" or "laryngeal" or "gland" or "oropharyngeal" or "salivary duct" or "base of tongue" or "posterior pharyngeal wall" or "tonsil" or "soft palate"), and ("cancer" or "tumor" or "carcinoma"). In addition to these terms, the references within the retrieved articles were examined to identify additional studies relevant to the assessment in the present research. Figure 1 shows the flow diagram of the research selection for relevant studies in this work.

| Inclusion criteria
Studies were included in this research if they met all of the following criteria. (a) The postoperative pathological diagnosis of each patient is HNSCC. (b) The CRP concentration in each patient's blood was evaluated before any treatment was administered. (c) The CRP pretreatment was measured using a serum-based method. (d) The associations between high CRP expression in the patient's blood and clinicopathological analysis, prognosis, or HNSCC status were demonstrated in this study. Studies were excluded from the present study based on the following criteria: (a) No relevant data could be extracted from the articles; (b) The study did not include survival outcomes or any useful HR with 95% confidence interval (CI); (c) Studies that contained no original experiments; (d) Studies that only reported elevated CRP or lowed CRP levels with no further explanation. In addition, to avoid pseudoreplication, when the data for the same patients appeared in more than one study and/or were combined with that of other patients, we only selected the most recent or the most complete study.

| Data extraction and quality assessment
The relevant data were extracted from the eligible studies by two independent researchers (R.C) and (YL.C) and examined The following relevant data were extracted: first author, year of publication, cancer type, country of patient population, gender, age, period of recurrence, median/mean age, follow-up period, sample size of patients, tumor pathology, cancer stage, cut-off value for CRP, and outcomes of patients, including overall survival (OS), cancer-specific survival (CSS), progress-free survival (PFS), and recurrence-free survival (RFS). The Newcastle-Ottawa Scale was likewise applied to assess the quality of the included papers.

| Statistical analysis
Statistical analysis was conducted using Stata software (12.0; Stata Corp LP). The relationships between CRP concentrations in the serum and the clinicopathological features, including gender of cases, tumor stage, and lymph node metastasis (LNM) of HNSCC cases, were evaluated by odds ratio (OR) along with 95% CI. For the prognostic effect of CRP, pooled HRs with 95% CI of CSS, RFS, OS, and PFS were used to evaluate the relationships between the CRP level survival outcomes of patients with HNSCC. We preferred to choose multivariate analyses when multivariate and univariate analyses of outcomes were available at the same time. If HR and 95% CI are not directly reported in the literature but Kaplan-Meier curves are provided, the HR with 95% CI can be directly obtained by extrapolating from the Kaplan-Meier curves. To evaluate the heterogeneity across the articles, we used the χ 2 and I 2 tests. p < 0.10 or I 2 > 50%, indicating significant heterogeneity, was considered optimal when using a random-effects model, but not a fixed-effects model, to estimate the pooled ORs/HRs. When p > 0.10 and I 2 < 50%, we preferred to use the fixed-effects model. In addition, to detect the stability of the present results, a one-way sensitivity analysis was performed by sequentially omitting each study included in this article. Begg test and funnel plots were adopted as well to evaluate any publication bias between the studies. All statistical tests were two-tailed, and p < 0.05 (95% confidence limit) was considered statistically significant.  24 Germany, 23,31 Japan, 19,22,29 Korea, 16 Switzerland, 25 Taiwan, 18 the United Kingdom, 26 and Canada. 32 The sample sizes of the included studies, which were retrospective and published between 2000 and 2020, ranged from 25 to 1589. The diverse treatments of patients with HNSCC included RT, CT, and CRT. The cut-off value of CRP was 1.2-10 mg/L in these studies. In the 17 articles, OS, PFS, CSS, and RFS were depicted in 14, 3, 3, and 3 studies, respectively.

| Publication bias and sensitivity analysis
A sensitivity analysis was applied to assess the contribution of each study to the estimations obtained from the pooled data. No obvious heterogeneity was observed, and sensitivity analysis again showed low heterogeneity in the included studies ( Figure 6A,B). The funnel plot of Begg's test exhibited symmetricalness with a P value of 0.462 in LNM analysis ( Figure 6C). Meanwhile, the funnel plot of Begg's test showed a P value of 0.743 in OS analysis ( Figure 6D). no publication bias was found in our study, indicating the statistically reliable results of the meta-analysis.

| DISCUSSION
Worldwide, the majority of new cases of patients with HNSCC annually are related to smoking and alcohol consumption. The five-year survival rate of patients with HNSCC is only approximately 50%, which is lower than those of other major carcinomas. 33,34 Sometimes, the behavior of the carcinoma and prognosis of the disease do not only depend on the clinical staging. 35 An evaluation of the clinical and histological parameters, as published in many studies, revealed that some less advanced cases of HNSCC had an early recurrence, whereas other advanced cases had good survival rates. [36][37][38] The TNM staging system is considered the most important prognostic factor for choice of therapy and prognosis, but it is still not accurate enough and should be supplemented with other indicators. Additionally, disease progression is associated with F I G U R E 5 Subgroup analysis of the association between CRP expression and OS in HNSCC (A) according to the cut-off value of CRP. (B) according to whether the value of HR was multivariate or univariate. (C) according to the tumor pathology. HNSCC: head and neck squamous cell carcinoma; CRP: C-reactive protein complex interactions between the tumor and the host's response to inflammation. The host's immune response releases a variety of mediators, which can promote chronic inflammation, cell proliferation, and even activation of unusual biochemical pathways. In tumor cases, these mediators eventually cause irreversible damage to the patient's DNA. 39,40 CRP is the first acute phase protein to increase, and within hours of the inflammatory response, it reaches concentrations thousands of times higher than normal. 41 CRP has a short half-life of only 6 h, and its concentration decreases rapidly when any tissue damage 42 is not found. In the USA, CRP is the second and third most important cause of cancer-related deaths among male and female patients, respectively. 43 Increased IL-6 is confirmed in patients with HNSCC. IL-6 is secreted by certain cells, including t lymphocytes, tumor cells, and peripheral blood mononuclear and normal cells. 44 Through the JAK-STAT signaling pathway, IL-6 regulates the expression of genes associated with and proliferation. 45 CRP is regulated by various cytokines like IL-6. 41 Elevated serum levels of IL-6 have been proven to be associated with elevated serum CRP levels. IL-6 can lead to inflammation and angiogenesis, which will subsequently raise the CRP levels. 44 Jing Yang et al. found that CD32/FcgRII on myeloma cells can bind to CRP, subsequently activating the downstream signaling through the transcription factor and p38 mitogen-activated protein kinase. This phenomenon results in increases in bone resorption and myeloma cell-mediated osteoclast differentiation in vivo caused by osteolytic cytokines produced by myeloma cells. 46 In Shasha Shen's study, CRP expression was upregulated in hepatocellular carcinoma (HCC) tissues compared with that in noncancerous tissues. The study demonstrated that the downregulation of CRP in HepG2 and Bel 7402 cells significantly inhibits cell growth, migration, and invasion in vitro, suggesting that high levels of CRP are carcinogenic in HCC. Moreover, it reported that the MEK/ERK and PI3K/AKT signaling pathways may be necessary for the CRP to stimulate cell migration and invasion of HCC cells. 47 In 2013, Lin's study reported that cases with colorectal cancer (CRC) had a poor prognosis when CRP levels were above 5 mg/L. CRP can act as a LOX-1 ligand in the CRC cell line. When the CRP concentration was 10 mg/L, the level of LOX-1 positive cells on the cell surface increased by almost 13%, resulting in an elevation of important genes in CRC cases. 48 These outcomes explained our results that elevated CRP levels are associated with LNM, high tumor stage, and worse prognosis, including CSS, PFS, and OS. In the subgroup analysis of CRP, when the cutoff value of CRP was ≥5 mg/L, a significant difference was observed in the correlation between CRP, LNM, and OS. As mentioned above, increased CRP is a symbol of poor prognosis. No association between tumor grade and CRP levels was found. However, this result should be treated with caution as it may be based on a small sample size available for use. In the subgroup analysis of multivariate HR value, significant difference was observed in the correlation between CRP and OS.
Moreover, males obtain worse prognosis in some tumors, including CRC and gastric cancer. The immunological, genetic, environmental, and hormonal differences in gender contribute to this unique situation. 49 Therefore, the high CRP in male patients with HNSCC may also be attributed to these gender differences. These outcomes explained our results that male patients are associated with elevated CRP levels, indicating a worse survival outcome.
There are some noteworthy limitations to our study. First, the type of cancer, staging, treatment strategy, and follow-up month of patients with HNSCC varied, all of which may have an impact on the pooled results. Second, the cut-off value of CRP varies from study to study, which might have led to heterogeneity. Third, our meta-analysis included only 17 studies. Thus, high-quality studies and a large sample size of studies are needed for verifying our results.

| CONCLUSION
Our study revealed high pretreatment levels of CRP indicated poor prognosis for HNSCC patients. CRP could easily be measured by doctors before treatment and can assist in predicting the prognoses of patients with HNSCC and guiding their treatments. Combining CRP and tumor stage enhanced the predictive power of CRP. In other words, CRP can be a valuable prognostic biomarker for patients with HNSCC. However, to verify our suggestions, additional prospective studies are needed that include large sample sizes, different methods, and different populations. In addition, if our results are verified, the determination of CRP levels should become standard practice for assessing patients with advanced head and neck carcinoma.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE
Not applicable.