Prognostic Role of Preoperative C-Reactive Protein in Upper Tract Urothelial Carcinoma after Radical Nephroureterectomy - An Updated Meta-Analysis

Background: C-reactive protein (CRP) is an acute-phase protein that is mainly stimulated by the inammation-associated cytokines and high concentrations of CRP have been detected in many tumors. Many studies have shown that there is a close correlation between CRP and the prognosis of upper urinary tract carcinoma (UTUC). However, the prognosis signicance of CRP in UTUC is still controversial. Our study aims to perform an updated systematic review and meta-analysis to investigate the association between CRP and the prognostic value for UTUC undergoing radical nephroureterectomy (RNU). Methods: A comprehensive electronic database search, using PubMed, EMBASE, and Web of Science, was performed to identify relevant research articles published before April 2020. Strict inclusion and exclusion criteria were established to extract data. Newcastle-Ottawa Scale was used to assess the quality of all candidate studies. The pooled hazard ratios (HR) and 95% condence intervals (CIs) for recurrence-free survival (RFS), cancer-specic survival (CCS), and overall survival (OS) were calculated to evaluate the intensity of association. And heterogeneity and publication bias were assessed. Moreover, to ensure the stability of the analysis results, a sensitivity analysis was tested. All data were analyzed by Stata 15.0 and Review Manager 5.3. Results: Thirteen studies including a total of 2801 patients were eligible for meta-analysis. Our results veried that elevated CRP predicted poorer FRS (pooled HR=1.38, 95%CIs (cid:0) 1.12-1.71, P=0.003), CCS (pooled HR=1.64, 95%CI (cid:0) 1.38-1.94, P<0.001), OS (pooled HR=1.35, 95%CIs: 1.03-1.79 ,P=0.03) in patients with UTUC after RNU. Besides, elevated CRP was overtly correlated with tumor pathological stage (T3/T4 vs. ≤ T2: OR=3.40, 95% CIs 2.39-4.84, P<0.001), tumor grade ((1/2 vs. 3: OR=1.68, 95% CIs 1.24-2.28, P<0.001), lymph node involvement (LNI) (+ vs. -: OR=2.73, 95% CIs 1.80-4.14, P<0.001), lymphovascular invasion (LVI) (+ vs. -: OR=4.1, 95% CIs 2.26-7.44, P<0.001). And these results got after using sensitivity analysis were veried to be stable and reliable. Furthermore, Begg's test result (RFS: P=0.089, CCS: P=0.592, and OS: P=0.089) showed no signicant publication bias in the procedure. Conclusion: Elevated CRP indicates a poor prognosis for patients with UTUC undergoing RNU. Elevated CRP plays a signicant role in reasonable risk stratication and individualized treatment. The present meta-analysis shows that elevated preoperative CRP is associated with poorer survival outcomes and aggressive clinicopathological features in UTUC after RNU. And the measurement of serum CRP levels is objective and reproducible. Hence, the test of preoperative CRP may be benecial for the treatment and prognostic evaluation of patients with UTUC. However, our results need more large-scale case-control studies and well-designed prospective studies to be conrmed.


Background
UTUCs are uncommon and account for only 5-10% of all urothelial carcinoma, [1] which is one of the most common tumors. About sixty percent of UTUCs are in the invasive stage at the time of diagnosis and patients aged 70-90 years have a peak incidence which can be observed. [1,2] Although a small number of low-grade patients with small lesions can be treated by endoscopic tumor ablation, for most patients, radical nephroureterectomy (RNU) with excision of an ipsilateral bladder cuff, including or not including retroperitoneal LN dissection, is the gold-standard treatment strategy for UTUCs. However, patients have a high rate of recurrence, including urinary bladder recurrence, local retroperitoneal recurrence and contralateral UTU, and poor prognosis after surgery. [1,3,4] CRP is an acute-phase protein that is mainly stimulated by the in ammation-associated cytokines, IL-6, during the acute-phase responses. Numerous causes can result in the changes in the plasma concentrations of CRP, including infectious and non-infectious factors, such as infection, trauma, surgery, burns, tissue infarction etc. [5] Moreover, many studies have found a strong association between elevated CRP and risk of cancers, such as hepatocarcinoma, lung cancer, skin cancer, renal cell carcinoma, bladder tumor, and lymphoma. [6][7][8] And preoperative CRP levels can predict disease characteristics, the extent of disease, and prognosticate survival in urologic cancers. [7] However, the exact effect of preoperative CRP in patients with UTUC is inconsistent in different studies. Although a previous meta-analysis study [9] has summarized that elevated CRP levels had an adverse prognostic value in UTUC after surgery, owing to a few enrolled studies, the accuracy of their results needed to be further con rmed. Therefore, we performed an updated systematic review and meta-analysis including more studies to perfect their research. At the same time, we got a dramatically different result that patients in an elevated CRP group had a poorer OS, which could predict worse outcomes in UTUC. And associations between abnormal CRP and clinicopathological features were also demonstrated.
In this study, we aimed to explore the prognostic value of elevated CRP levels for RFS, CCS, and OS in patients with UTUC by the method of pooling outcomes (HR and 95%CIs) from available data and associations between elevated preoperative CRP and part of clinicopathological features.

Literature search strategy
According to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, a comprehensive electronic databases search by employing PubMed, EMBASE, and Web of Science was performed to identify relevant research articles which were published before April 2020. To ensure comprehensive retrieval, Medical Subject Headings with free words in the title/abstract were combined to perform the research strategy. Our research terms included "Upper urinary tract" OR "Upper tract" OR "Kidney Pelvis" OR "Ureter" AND "Carcinoma, Transitional Cell" OR "Urothelial carcinoma" AND "CRP" OR "Protein, C-Reactive" AND "Prognosis" OR "Predictor" OR "Cohort studies". All studies written in English were included.

Study inclusion criteria and relevant de nitions
All studies meet the following restrict inclusion criteria: (1) the diagnosis of UTUC must be veri ed by postoperative pathological examination; (2) all UTUC patients had undergone RNU; (3) preoperative measurement of serum CRP was routinely performed up to a week prior to RNU; (4) studies reported HRs and 95% CIs for preoperative serum CRP level in RFS, and(or) CCS, and(or) OS or raw data contained in articles, from which outcomes can be calculated; (5) the study design belonged to a prospective or retrospective cohort study.
When disease recurrence was de ned as local recurrence in the operative site, regional lymph nodes, or distant metastasis, RFS was calculated from the time of curative treatment to tumor recurrence. And CSS was de ned as the time from the date of diagnosis to the date of death caused by UTUC. In addition, OS was measured from the date of de nite diagnosis to death or last follow-up of patients.

Data extraction
The retrieved studies were carefully screened and meaningful data were extracted by two independent researchers(XY Qian and ZX Wang). If there were different opinions about relevant information, they discussed them with the third researcher (C Qian) to come to an agreement. By browsing the title and abstract, the irrelevant studies were excluded. The full texts were then read to enroll quali ed studies, according to the set inclusion and exclusion criteria of our meta-analysis. The data we extracted include (1) basic characteristics for research including rst author's surname, year of publication, study region, sample size; (2) patients' information including gender, age, follow-up time; (3) data of UTUC including size, stage, tumor grade, LN (lymph node) status, LVI (lymphovascular invasion) status, and the cut-off value of preoperative serum CRP; (4) outcomes of interest including HR and 95% CIs for RFS, CSS, and OS.
For studies sharing an identical population, we only chose the most informative or recent study for further analyses. Only multivariate outcomes were extracted if univariate and multivariate outcomes were provided. However, univariate outcomes were not acceptable if multivariate results weren't obtained in our study.

Quality assessment
Quality assessment of enrolled studies was evaluated by two independent researchers(XY Qian and ZX Wang), using the Newcastle-Ottawa quality assessment scale (NOS) [10] which ranges from 0 to 9 stars. Only if the study score was greater than or equal to 6 stars, could this study be graded as high quality, which was included in this meta-analysis.

Statistical analysis
Outcomes of interest with HR along with 95%CIs were pooled to evaluate the in uence of preoperative serum CRP on RFS, CSS, and OS. Statistic heterogeneity among included studies was assessed by the Cochrane Q statistic and I 2 statistic. And P heterogeneity <0.1 or I 2 >50% meant that heterogeneity existed among studies and that we should use the random-effects model to pool outcomes of interest. However, if heterogeneity didn't exist, the xed-effects model would be chosen. And sensitivity analysis was performed to assess the robustness of the results by eliminating study one by one. Furthermore, publication bias was also examined by a funnel plot and Begg's test. The whole process of the meta-analysis was performed with Stata 15.0 and Review Manager 5.3. And all statistical tests were two-sided and the statistical signi cance level was set at P<0.05.

Search results
A total of 392 publications were retrieved by using our initial search algorithm we had developed. There were 312 records left after removing duplicate articles. And 262 articles were excluded by screening the title and abstract. The remaining 50 articles were read in full text, and it is 13 studies [11][12][13][14][15][16][17][18][19][20][21][22][23] that satis ed the inclusion criteria and were analyzed in our meta-analysis. The speci c literature screening process was displayed in Figure 1.

Study characteristics
The clinical characteristics of all 13 eligible publications were summarized in Table1. All cohorts enrolled about 2801 patients with median or mean age ranging from 65.7 to 74 years and belonged to a retrospective study, which had the median duration of follow-up varying from 15.1 to 56.4 months. Of the 13 studies, 9 were reported in Japan, 2 in German, 1 in Korea, and 1 in multi-centers which was conducted in Japan. And the serum CRP cut off ranged from 0.13mg/dl to 1.0mg/dl, which were mainly at 0.3 and 0.5 mg/dl. All patients didn't nd distant metastasis before surgery. The enrolled patients all were treated by RNU, and the main pathologic features of the included patients are shown in Table 2

Elevated preoperative CRP and FRS in UTUC
There were 5 cohorts extracted the data of preoperative CRP and FRS in patients with UTUC. Meta-analysis results of the association between elevated CRP and RFS of UTUC pointed out that the signi cant heterogeneity (I 2 = 68%; P heterogeneity = 0.01) between enrolled cohorts was detected. Therefore, a sensitivity analysis was conducted, whose result suggested that Obata J and his colleagues' study [19] was the main reason causing the heterogeneity. Deleting this study, we performed the meta-analysis again by using the xed-effect model and got perfect results that there was no signi cant heterogeneity(I 2 = 0%; P heterogeneity = 0.85) between remaining 4 studies and that elevated CRP was strongly correlated with short FRS (HR=1.38; 95%CIs: 1.12-1.71; P=0.003)( Figure  2a).

Elevated preoperative CRP and CCS in UTUC
After the data of 12 cohorts about HR and 95%CIs for CCS was dealt with by the random-effect model, there was a signi cant heterogeneity presented of this heterogeneity. Removing two studies, non-signi cant heterogeneity (I 2 = 0%; P heterogeneity = 0.45) were found and we chose the xed-effects model to pooled the HR and 95%CIs of left 10 articles. Besides, the pooled results indicated that the more adverse CCS occurred in patients with UTUC, accompanying elevated CRP (HR=1.64; 95%CIs: 1.38-1.94; P<0.001) (Figure 2b). What's more, although all included studies used inconsistent cut-off values for CRP, it wasn't a source causing signi cant heterogeneity, which was demonstrated by a sensitivity analysis.

Preoperative CRP and OS in UTUC
When the association between preoperative CRP and OS was explored, the data of 4 (Total:4) cohorts were calculated. The signi cant heterogeneity wasn't found and the xed effect model was considered (I 2 =9%; P heterogeneity =0.35). That patients, who were diagnosed with UTUC, with elevated preoperative CRP had a poorer OS was seen (HR=1.35; 95%CI: 1.03-1.79; P=0.03) (Figure 2c).

Preoperative CRP and clinicopathological features in UTUC
The above procedures were repeated to merge the relevant results. And the nal meta-analysis ndings were that elevated preoperative CRP was overtly correlated with tumor pathological stage (III/IV  All pooled results were presented in Figure 3.

Publication bias
To investigate whether publication bias existed in results, the funnel plots were depicted respectively (Figure 4). There was no signi cant publication bias that existed in FRS, CCS, and OS by Begg's test (P =0.089, P =0.592, P = 0.089, respectively).

Discussion
Our study aimed to explore the prognostic value of preoperative elevated CRP levels for patients with UTUC after RNU. In our analysis, 13 cohort studies including 2801 UTUC patients were included to pool the outcomes and our results displayed the associations between abnormal CRP and prognosis of UTUCs. Data we calculated shown that, in patients with UTUC receiving RNU, high preoperative CRP predicted poorer FRS (pooled HR = 1.38, 95%CIs: 1.12-1.71, P = 0.003), CCS (pooled HR = 1.64, 95%CIs 1.38-1.94, P < 0.001), and OS (pooled HR = 1.35, 95%CIs:1.03-1.79, P = 0.03). And there was no signi cant publication bias which existed in FRS, CCS and OS, respectively (Begg's test: P = 0.089, P = 0.592, P = 0.089). Compared with You L and his colleagues' study [9], we recon rmed that elevated preoperative CRP had a poor impact on RFS and CCS of patients with UTUC and found elevated CRP also played a signi cant adverse effect on OS in patients with UTUC treated by RNU, which greatly differed from theirs. On the basis of their research, in which there were 2 studies [3,24] only containing outcomes in the univariate analysis, we retrieved another 8 relevant studies [12][13][14][15][16][17][18][19] including sound results in multivariate analysis, and, since then, 1233 cases were added to our studies. What's more, we found that authors neglecting the heterogeneity between all studies used a random-effects model to pool all outcomes of interest in their analysis, which may have a certain in uence on the accuracy of the nal results. Therefore, this updated metaanalysis increased the number of UTUC cases up to 2801 and incorporated more reliable multivariate analysis results. And we made up for their improper method used in their procedure and perfected their relevant research.
Except for other causes, increased serum CRP levels as a biomarker often associated with malignancy in cancer patients. In UTUCs, a high CRP value (> 0.5 mg/dl or 0.9 > mg/dl) was signi cantly associated with progressive tumor characteristics, including muscle-invasive disease (pT ≥ 2), high tumor stage, high grade, lymph node metastasis, positive LVI, distant metastasis, poor tumor differentiation, etc [3,11,19,20], which was consistent with what we pooled. And detecting serum CRP concentration is objective and reproducible measurement before the operation. Therefore, it may easily be implemented as a prognostic factor for the degree of malignancy and to more accurately stratify patients with UTUC before surgery, which could guide treatment strategy, but more prospective and retrospective studies are needed to con rm this result.
Although the molecular mechanism of abnormally elevated CRP in patients diagnosed with cancer in various studies remains unclear, there were several plausible explanations why CRP could be elevated. One of these possible mechanisms is that the increasing serum CRP as one of many acute-phase proteins which were mainly induced by in ammation-associated cytokines [5] -interleukin-6, interleukin-1β, tumor necrosis factor α, interferon-γ, transforming growth factor β, and interleukin-8 -re ects the in ammatory state of the body caused by tumors, which may contribute to tumor cell more rapid growth [25,26].
Alternatively, immune suppression caused by changes in tumor microenvironment resulted in sustaining tumor cell growth and activating the capability for tumor invasion and metastasis. [27] Generally speaking, proin ammatory cytokines and immune dysfunction signi cantly affect the serum CRP levels and indirectly display the growth capacity of tumors. And elevated CRP levels has great reference value for the prognosis of cancer patients. According to our pooled data, it was RFS, CCS, and OS that elevated CRP had a great in uence on. And in our study, that patients with UTUC undergoing RNU had a poor OS in the elevated CRP groups shown statistical signi cance, which also demonstrated more adverse prognosis in patients diagnosed with UTUC.
While our meta-analysis offers plenty of evidence that increased preoperative CRP is closely bound up with more adverse prognosis of patients with UTUC treated by RNU, some limitations in our study can be detected, which should be acknowledged. Firstly, signi cant heterogeneity existed for RFS and CCS. After heterogeneity in studies was determined and sensitivity analyses were conducted, there was some mild heterogeneity in included studies. Secondly, limited databases (Web of Science, PubMed, and EMBASE) were searched, which might make some valid articles missed and have the estimating power limited. Thirdly, all of the enrolled studies were retrospective and published in English, most of which published by Japanese scholars. Therefore, information and selection biases may exist in studies. And publication bias is more inclined to be discovery, owing to the fact that present studies including positive results are more easily to be published in English journals.

Conclusion
The present meta-analysis shows that elevated preoperative CRP is associated with poorer survival outcomes and aggressive clinicopathological features in UTUC after RNU. And the measurement of serum CRP levels is objective and reproducible. Hence, the test of preoperative CRP may be bene cial for the treatment and prognostic evaluation of patients with UTUC. However, our results need more large-scale case-control studies and well-designed prospective studies to be con rmed.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analysed during this study are included in this published article [and its supplementary information les].

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable.

No. Author
Year Study Flowchart of the search and selection process in this meta-analysis. Funnel plots for publication bias of the hazard ratios (HRs) of (b) RFS, (b) CSS, and (c) OS.