Invasiveness of Upper Tract Urothelial Carcinoma: Clinical Significance and Integrative Diagnostic Strategy
Article information
Abstract
Purpose
In this study, we aimed to determine the clinicopathologic, radiologic, and molecular significance of the tumor invasiveness to further stratify the patients with high-grade (HG) upper tract urothelial carcinoma (UTUC) who can be treated less aggressively.
Materials and Methods
Clinicopathologic and radiologic characteristics of 166 surgically resected HG UTUC (48 noninvasive, and 118 invasive) cases were evaluated. Six noninvasive UTUC cases with intratumoral tumor grade heterogeneity were selected for whole-exome sequencing (WES) to understand the underlying molecular pathophysiology. Barcode-tagging sequencing was done for validation of the target genes from WES data.
Results
Patients with noninvasive UTUC showed no cancer-specific death with better cancer-specific survival (p < 0.001) and recurrence-free survival (p < 0.001) compared to the patients with invasive UTUC. Compared to the invasive UTUC, noninvasive UTUC was correlated to a low grade (LG) on the preoperative abdominal computed tomography (CT) grading system (p < 0.001), histologic intratumoral tumor grade heterogeneity (p=0.018), discrepancy in preoperative urine cytology diagnosis (p=0.018), and absence of urothelial carcinoma in situ (p < 0.001). WES of the heterogeneous components showed mutually shared HRAS and FGFR3 mutations shared between the HG and LG components. HRAS mutation was associated with the lower grade on preoperative abdominal CT and intratumoral tumor grade heterogeneity (p=0.045 and p < 0.001, respectively), whereas FGFR3 mutation was correlated to the absence of carcinoma in situ (p < 0.001).
Conclusion
According to our comprehensive analysis, HG noninvasive UTUC can be preoperatively suspected based on distinct preoperative radiologic, cytologic, histologic, and molecular features. Noninvasive HG UTUC shows excellent prognosis and thus should be treated less aggressively.
Introduction
Upper tract urothelial carcinoma (UTUC) comprises a tumor arising from the urothelium of the renal calyces to the ureteric orifices. UTUC is rare compared to urothelial carcinoma of the bladder, accounting for only 5%-10% of all urothelial carcinomas and < 10% of renal tumors [1]. Accurate grading and staging is often only available after radical nephroureterectomy due to the intratumoral heterogeneity, the difficult location to approach using the current endoscopic biopsy techniques, and limited accuracy of imaging technologies. Thus, radical nephroureterectomy and excision of the bladder cuff remains the gold standard for the management of UTUC. However, it increases the burden for the patient predisposed to chronic kidney disease or end-stage renal failure requiring dialysis and possibly leads to secondary complications such as hypertension, diabetes, and obesity in the long term [2].
Recent studies have demonstrated the possibility to preserve the ipsilateral kidney via kidney-sparing surgery or endoscopic laser resection with similar oncological and survival outcomes in case of low-risk UTUC [1,3,4]. Low-risk UTUC is restricted to only a limited number of tumors with the following features: unifocality, tumor size less than 2 cm, no invasive focus in preoperative imaging, and low-grade (LG) histology [2]. Ongoing studies using preoperative imaging are aimed to predict the pathologic T (pT) category of UTUC. Honda et al. [5,6] showed a high diagnostic accuracy in discriminating tumors invading beyond muscularis (≥ pT3) from the others (≤ pT2). However, there were no successful image studies that discriminated noninvasive UC from the invasive UC. UTUC is usually preoperatively diagnosed via ureteroscopic biopsy or cytology. An accurate grading of the tumor is also difficult in cases with histologic intratumoral tumor grade heterogeneity, and the clinicopathologic significance of the intratumoral tumor grade heterogeneity is not yet known. Consequently, due to the limitations of the preoperative image and biopsy/cytology results, radical nephroureterectomy is performed for full examination of most patients with UTUC. However, considering risk factors such as old age, bilateral UTUC, or single kidney, this is a major concern and a difficult decision for clinicians. Therefore, patients with UTUC must be stratified according to tumor invasiveness and/or tumor grade.
With the exponential advances in molecular studies, research efforts are focused on understanding the pathophysiology and underlying molecular mechanism of UTUC. Sfakianos et al. [7] conducted the largest targeted genomic study of UTUC with 300 cancer-associated genes and showed that the prevalence of FGFR3, HRAS, and CDKN2B alteration is higher in UTUC than in bladder cancer [7]. By a whole-exome sequencing (WES) of DNA and RNA and protein analyses of UTUC, Moss et al. [8] showed that FGFR3 is the most commonly mutated gene in both the LG (92%) and the high grade (HG) (60%) tumors. Fujii et al. [9] recently proposed five molecular subtypes of UTUC and their clinicopathologic features. However, previous molecular studies of UTUC evaluated only a few cases or incomplete clinicopathologic data especially regarding histologic tumor grade and/or invasiveness, while most UTUC studies rather had a heterogeneous cohort of mostly invasive UTUC cases.
In the present study, we aimed to determine the prognostic significance of UTUC by centrally reviewing 166 HG noninvasive and invasive UTUC cases and assessing the clinicoradiologic, pathologic, and molecular characteristics with a focus on the significance of invasiveness. We also attempt to understand the underlying molecular pathophysiology of the histologic intratumoral tumor grade heterogeneity in order to identify the factors that can be used to further stratify patients with UTUC.
Materials and Methods
1. Patient selection
This study was approved by the Institutional Review Board (approval number: S2021-0705-0001). A total of 174 cases of surgically resected HG UTUC at Asan Medical Center, Seoul, Republic of Korea, from 2016 to 2020 were retrieved. Three cases were excluded: one case of neoadjuvant chemotherapy and two cases with unavailable slides or formalin-fixed paraffin-embedded (FFPE) blocks. After a central pathologic review according to the 2022 World Health Organization (WHO) classification [10], one case of noninvasive UTUC were re-categorized as invasive UTUC and five other cases as LG, which were also excluded from this study. Finally, a total of 48 HG noninvasive UTUC and 118 HG invasive UTUC cases were analyzed. A flowchart of patient selection criteria is depicted in S1 Fig. Clinical data, including age, sex, procedure, presence of distant metastasis, and survival outcomes, were obtained from electronic medical records. Cancer-specific survival (CSS) was defined as the time from the day of radical nephroureterectomy to the death due to UTUC. Recurrence-free survival (RFS) was defined as the time from the day of radical nephroureterectomy to local or distant recurrence. Patients with prior bladder cancer history were also included in this study, therefore, intravesical, urethral, ipsilateral or contralateral pelvic or ureteral recurrence were not considered as recurrence [11].
2. Radiologic assessment
A genitourinary-specialized radiologist with 7 years of experience in genitourinary imaging evaluated the preoperative computed tomography (CT) images according to CT grading system proposed by Honda et al. [6]. Grade A lesions are characterized by symmetric wall thickening without and with spiculation or masses with smooth surface without spiculation (Fig. 1A and B). Grade B lesions are defined by asymmetric circumferential ureteral wall thickening with spiculation, masses with smooth surface and spiculation, and masses with irregular surface with or without spiculation (Fig. 1C and D) [6].
Representative abdominal computed tomographic (CT) images of high-grade upper tract urothelial carcinoma (UTUC). Axial (A) and coronal (B) views of the CT scan shows 2-cm tumor (arrows) without spiculation in the right distal ureter, causing hydroureterosis. This lesion was radiologically classified as grade A (pattern 3) and was pathologically confirmed as noninvasive high-grade UTUC. Axial (C) and coronal (D) reformatted CT scans reveal a 2.2-cm mass with irregular surface (arrows) in the left distal ureter and ureterovesical junction. This lesion was radiologically classified as grade B (pattern 6) and was pathologically confirmed as pT2.
3. Histologic evaluation
Pathologic data, including tumor location and size, were retrieved from surgical pathology reports. All hematoxylin and eosin–stained slides were carefully re-evaluated by two genitourinary pathologists who were both blinded to the clinicopathologic information. Tumor grade was according to the 2022 WHO classification by a cutoff of 5%, in which pure LG tumor or LG tumor containing less than 5% of the HG component was classified as LG and hence excluded from the study [10]. Histologic intratumoral tumor grade heterogeneity was defined as the presence of both definite HG and LG components but with the LG component at more than 10%. In cases with intratumoral tumor grade heterogeneity, the tumor was scanned at a low power view (Fig. 2A), and the typical area of HG and LG components were assessed and manually marked. LG component was defined as an area with none of the characteristics of HG component, but with an orderly architecture and uniform tumor cell size (Fig. 2A, right lower inlet). HG component was defined as an area with nuclear pleomorphism, hyperchromasia, high nuclear/cytoplasm ratio with frequent mitotic figures, or necrosis (Fig. 2A, left lower inlet). Preoperative biopsies, if present, were also re-evaluated (Fig. 2B). Invasion depth, lymphovascular invasion (LVI), perineural invasion (PNI), resection margin status, presence of urothelial carcinoma in situ, and lymph node metastasis were assessed.
Representative microscopic images of a case of noninvasive upper tract urothelial carcinoma with intratumoral heterogeneity. (A) Papillary structure of neoplastic urothelium of variable thickness is noted with heterogenous components of both high grade and low grade (H&E, ×1.0). Low-grade component shows a relatively orderly appearance with only mild nuclear atypia (H&E, ×40, lower right inlet). High-grade component shows cellular disorder, nuclear size variation and pleomorphic nuclei with frequent mitosis (H&E, ×40, lower left inlet). (B) Preoperative biopsy shows a low-grade noninvasive urothelial carcinoma with delicate papillae of urothelial proliferation with mild nuclear irregularity.
4. WES of the intratumoral heterogeneous components of HG noninvasive UTUC
In order to compare the mutation profiles of each HG and LG components within the tumor, HG, noninvasive UTUC cases with discrepant preoperative biopsy result as LG, representing cases with most intratumoral tumor grade heterogeneity, were selected for WES. Six cases were identified, from which whole tumor sections were carefully reviewed by two pathologists and typical areas of HG and LG tumor components and normal renal parenchyma were representatively marked multifocally on the slide of the resected specimen. Each heterogeneous area and the preoperative biopsy specimens were manually and separately macrodissected and subjected for WES (S2 Fig.).
To generate standard exome capture libraries, we used the Agilent SureSelect Target Enrichment protocol for Illumina paired-end sequencing library (ver. C2, December 2018) together with 1 μg input genomic DNA and 200 ng input FFPE DNA. SureSelect Human All Exon V6 probe set was used in all cases. DNA was quantified and the DNA quality was measured using PicoGreen and agarose gel electrophoresis. We used 200 ng of FFPE DNA diluted in EB buffer and sheared to a target peak size of 150-200 bp using the Covaris LE220 focused ultrasonicator (Covaris, Woburn, MA) according to the manufacturer’s recommendations. We loaded the 8-micro TUBE strip into the tube holder of the ultrasonicator and sheared the DNA using the following settings: mode, frequency sweeping; duty cycle, 10%; intensity, 5; cycles per burst, 200; duration, 60 seconds×6 cycles; and temperature, 4-7°C. The fragmented DNA was repaired, an “A” was ligated to the 3′ end, and Agilent adapters were then ligated to the fragments. Upon assessment of the ligation, the adapterligated product was polymerase chain reaction (PCR) amplified. For exome capture, 250 ng of DNA library was mixed with hybridization buffers, blocking mixes, RNase block, and 5 μL of SureSelect all exon capture library, according to the standard Agilent SureSelect Target Enrichment protocol. Hybridization to the capture baits was conducted at 65°C using a heated thermal cycler lid option at 105°C for 24 hours on a PCR machine. The captured DNA was then washed and amplified. The final purified products were then quantified using quantitative PCR (qPCR) according to the qPCR Quantification Protocol Guide (KAPA Library Quantification kits for Illumina Sequencing platforms) and qualified using the TapeStation DNA screentape D1000 (Agilent Technologies, Waldbronn, Germany). Sequencing was performed using the HiSeq 2500 platform (Illumina, San Diego, CA).
Sequenced reads were aligned to the human reference genome (GRCh38) with BWA (v0.5.9) with default options [12]. MarkDuplicates of Picard package (v2.5.0) (http://broadinstitute.github.io/picard) were used to remove PCR duplicates from the aligned reads. GATK IndelRealigner (1.6.5) was used to realign deduplicated reads at known indel positions. GATK Table Recalibration was used to recalibrate the base quality [13]. Mutect (v1.1.7) was used to detect somatic single nucleotide variants and indels [14].
5. Validation of HRAS and FGFR3 mutation by barcode-tagged sequencing
After excluding six patients who underwent WES, a total of 42 HG noninvasive UTUC patients and 118 HG invasive UTUC patients were subjected for validation to examine hotspot mutations in HRAS (12, 13, and 61 amino acids) and FGFR3 mutation (248, 249, 370, and 373 amino acids); these two were mutually shared genes between the LG and HG components of UTUC on the WES. Excluding the inflammatory cells or necrosis, the tumor area was marked and macrodissected in each patient, and DNA was extracted from the FFPE blocks using QIAamp DNA FFPE Tissue Kit. Exon 2 and 3 of HRAS and exon 7 and 10 of FGFR3 were amplified using the constructed primers (S3 Table). QIAquick Gel Extraction Kit was used to purify the amplicons, which were then sequenced using in-house designed primer sets and Illumina platform-based barcode-tagging sequencing (BTseq) whole genome sequencing kit (Celmics, Seoul, Korea) for multiplex amplicon sequencing on a MiSeq sequencer (150 bp paired-end more, Illumina).
6. Data acquisition from The Cancer Genome Atlas
Mutation profiles of UTUC of The Cancer Genome Atlas (TCGA) data were obtained from cBioPortal [15].
7. Statistical analysis
Mann-Whitney U test was conducted to evaluate the differences between continuous variables in two groups, whereas Kruskal-Wallis test was employed to assess the differences in three or more groups. Chi-square test and Fisher’s exact test were used to determine the differences in the frequencies of categorical variables and very small expected frequencies, respectively. Univariate analysis of factors associated with survival was assessed using the Kaplan-Meier method and compared by the log-rank test. Cox proportional hazards regression model was used for multivariate survival analysis of clinicopathologic factors significantly associated with survival on univariate analyses. The trends of relative hazard according to the continuous variables were estimated using both Martingale residual of null Cox model [16] and univariate Cox model using a restrictive cubic spline-transformed variable [17]. All statistical evaluations were performed using R ver. 4.0.3 (The R Foundation for Statistical Computing, Vienna, Austria), with p-values < 0.05 considered statistically significant.
Results
1. Patient characteristics of noninvasive and invasive HG UTUC
The clinicopathologic features of 48 patients diagnosed with HG noninvasive UTUC and 118 patients with HG invasive UTUC are summarized in Table 1. All patients underwent radical nephroureterectomy regardless of tumor invasiveness.
Clinicopathologic characteristics of invasive and noninvasive, high grade upper tract urothelial carcinoma
The median follow-up period was 61.5 months, during which five patients (10.4%) with noninvasive UTUC died, but none of them were cancer-specific deaths. In contrast, 29 patients (24.6%) with invasive UTUC had cancer-specific deaths. CSS was significantly better in patients with noninvasive UTUC (5-year survival rate, 100%) than in those with invasive UTUC (52.9%, p < 0.001) (Fig. 3A). Similarly, RFS was significantly better in patients with noninvasive UTUC (5-year survival rate, 95.5%) than in those with invasive UTUC (54.8%, p < 0.001) (Fig. 3B).
Cancer-specific survival (CSS) and recurrence-free survival (RFS) of patients with high-grade upper tract urothelial carcinoma (UTUC). (A) Patients with noninvasive UTUC showed a significantly better CSS than invasive UTUC (5-year survival rate, 100% vs. 52.9%; p < 0.001). (B) Patients with noninvasive UTUC showed a significantly better RFS than invasive UTUC (5-year survival rate, 95.5% vs. 54.8%; p < 0.001).
Tumors were classified as grade A more frequently among patients with noninvasive UTUC (n=44, 91.7%) than among those with invasive UTUC (n=14, 11.9%; p < 0.001) according to the abdominal CT grading system. Urothelial carcinoma in situ was detected significantly less frequently among patients with noninvasive UTUC than those with invasive UTUC (p < 0.001). On the other hand, histologic intratumoral tumor grade heterogeneity (p=0.018) was frequently observed in patients with noninvasive UTUC compared to those with invasive UTUC. Discrepancy in preoperative urine cytology was higher in patients with noninvasive UTUC (p=0.018). Preoperative biopsy was completed in 20 (41.7%) and 51 (43.2%) patients with noninvasive and invasive UTUC, respectively. Patients with noninvasive UTUC showed a higher discrepancy rate in preoperative biopsy specimens (30.0% vs. 13.5%), but without statistical significance (p=0.196). In addition, discrepancy between preoperative biopsy and surgical specimens were significantly correlated to intratumoral tumor grade heterogeneity (p=0.040) (S4 Fig.).
2. WES results
Since grade discrepancy rates of preoperative biopsy specimens were higher in patients with noninvasive UTUC than in those with invasive UTUC, and since all of these cases accompanied intratumoral tumor grade heterogeneity, we next planned to understand the underlying molecular alterations in each of the HG and LG components. Six cases of HG noninvasive UTUC showed discrepant preoperative biopsy results similar to LG, but two cases were excluded from further analysis due to poor DNA quality and inadequate specimen. Therefore, only four HG noninvasive UTUC cases underwent WES. The results are depicted as a phylogenetic tree in Fig. 4. The phylogenetic tree shows a clonal evolution starting from the LG (both the resection specimen and the biopsy specimen) to the HG components. Among the several shared mutations between the HG and LG components, HRAS and FGFR3 mutations were noted in three and one cases, respectively (Fig. 4). This finding was remarkable since both mutations are conventionally thought to be LG markers, however were retained throughout the HG transformation without discernable gene mutations of p53 and/or Rb, which are characteristically HG markers. Therefore, we hypothesized that HRAS and FGFR3 may have clinicopathologic impact in UTUC, and may be responsible for the frequent intratumoral tumor grade histologic heterogeneity of noninvasive UTUC.
Phylogenetic tree of four high-grade, noninvasive upper tract urothelial carcinoma cases. Whole-exome sequencing was done for four cases with intratumoral heterogeneity. Each of the low-grade (LG) and high-grade (HG) components was manually dissected for mutational analysis. Each branch of the phylogenetic tree designates each component. FGFR3 mutation and HRAS mutations were shared between the LG and HG components of case 1 and cases 2-4, respectively.
3. Clinicopathologic significance and survival outcome of HRAS and FGFR3 mutation in HG UTUC patients
To validate the WES results, 116 cases in the validation cohort were examined for HRAS (Q61L, Q61R, and Q61H) and FGFR3 (M373V, R248C, and S249C) hotspot mutations. HRAS and FGFR3 alterations were noted in 58 (34.9%) and 21 (13.0%) cases, respectively, in the total of 166 and 162 HG UTUC cases by BTseq, respectively (four cases were not available for FGFR3 analysis due to scanty tissue). Correlation of HRAS and FGFR3 alterations with other clinicopathologic factors are summarized in S5 and S6 Tables, respectively. In detail, the alteration of HRAS mutation, although marginally significant (p=0.058), was more frequent in the invasive UTUC cases than in the noninvasive UTUC cases. The presence of HRAS mutation was correlated with tumors located in the kidney (p=0.012), a higher radiologic grade (p=0.049), tumors with higher pathologic T category (p=0.034), a higher American Joint Committee on Cancer (AJCC) stage (p=0.045), and an intratumoral tumor grade heterogeneity (p < 0.001) (Fig. 5A). The presence of FGFR3 mutation was significantly correlated with the absence of urothelial carcinoma in situ (p < 0.001) (Fig. 5B). CSS and RFS were not significantly different between patients with HRAS mutation (5-year survival rate, 66.7% and 59.2%, respectively) and those without HRAS mutation (70.4%, p=0.335; 70.4%; p=0.160, respectively) (Fig. 5C). Patients with FGFR3 mutation showed a tendency of better survival but did not meet the statistical significance in CSS from those without FGFR3 mutation (5-year survival rate, 95.0% vs. 65.2%; p=0.200). However, patients with FGFR3 mutation showed significantly better RFS than those without HRAS mutation (5-year survival rate, 95.2% vs. 61.7%; p=0.044) (Fig. 5D).
Clinicopathologic significance and survival impact of HRAS and FGFR3 mutations. (A) Presence of HRAS mutation was significantly correlated with tumor located in the kidney (p=0.012), tumor with higher pathologic T category (p=0.034), higher American Joint Committee on Cancer (AJCC) stage (p=0.045), and intratumoral heterogeneity (p < 0.001). (B) Presence of FGFR3 mutation was significantly correlated with the absence of urothelial carcinoma in situ (p < 0.001). (C) Cancer-specific survival (CSS; 5-year survival rate, 66.7% vs. 70.4%; p=0.335) and recurrence-free survival (RFS; 5-year survival rate, 59.2% vs. 70.4%; p=0.160) were not significantly different between patients with HRAS mutation and those without HRAS mutation. (D) Patients with FGFR3 mutation showed a tendency of better survival but did not show any significant difference in CSS from those without FGFR3 mutation (95.0% vs. 65.2%, p=0.200). However, patients with FGFR3 mutation showed significantly better RFS than those without HRAS mutation (95.2% vs. 61.7%, p=0.044).
4. Univariate and multivariate analyses of the RFS in HG UTUC patients
Since the urinary tract recurrence was not considered as recurrence, cases with positive resection margin (n=3) were excluded from analysis of the RFS. On univariate analysis, poor RFS was caused by a higher radiologic grade B (p=0.002), pT category (p < 0.001), and AJCC stage (p < 0.001), as well as the presence of lymph node metastasis (p < 0.001), preoperative distant metastasis (p < 0.001), LVI (p < 0.001), PNI (p <0.001), and urothelial carcinoma in situ (p=0.007) (Table 2).
Univariate and multivariate analysis of the clinicopathologic factors on recurrence-free survival (n=163)
Multivariate analyses were performed to determine whether these clinicopathologic factors remain a predictor of RFS after adjusting for these factors. As T classification, lymph node metastasis, and distant metastasis are components of the AJCC stage group, the three components, instead of the AJCC stage group, were separately assessed in the multivariate analysis. Results showed that lymph node metastasis (p=0.006), preoperative distant metastasis (p=0.001), presence of LVI (p < 0.001), and PNI (p < 0.001) were independent poor prognostic factors of HG UTUC (Table 2).
5. Univariate and multivariate analyses of the CSS in HG UTUC patients
On univariate analysis, poor CSS was caused by older age (p=0.047), higher radiologic grade (p=0.005), higher pT category (p < 0.001), lymph node metastasis (p=0.002), distant metastasis (p < 0.001), higher AJCC stage (p < 0.001), presence of LVI (p < 0.001), PNI (p=0.025), urothelial carcinoma in situ (p=0.010), and involvement of resection margin (p < 0.001) (Table 3).
Univariate and multivariate analysis of the clinicopathologic factors on cancer-specific survival (n=166)
Based on multivariate analysis, older age (p=0.044), presence of LVI (p=0.028), and involvement of resection margin (p=0.014) were independent poor prognostic factors of HG UTUC (Table 3).
6. TCGA analysis
Among 227 UTUC cases (137 invasive and 44 noninvasive) in TCGA data, only 118 cases had clinical data on tumor invasiveness. The overall mutation landscape of these cases showed most frequent alterations in FGFR3 (40%), KMT2D (36%), KDM6A (30%), TP53 (28%), and ARID1A (25%) (S7A Fig.). The differential gene mutations between the invasive and noninvasive UTUC with a p-value of < 0.05 were noted in NOTCH2 (13.9% vs. 0%), NOTCH4 (10.2% vs. 0%), BRCA1 (10.9% vs. 0%), EPHB1 (9.5% vs. 0%), DNMT1 (8.8% vs. 0%), TP53 (36.5% vs. 2.3%), FGFR3 (32.1% vs. 63.6%), and STAG2 (11.7% vs. 31.8%) (S7B Fig.). HRAS mutation was present in 9.5% (13/137) and 11.4% (5/44) of invasive and noninvasive UTUC, but not statistically significant.
Discussion
In the present study, a central review with a thorough pathologic, clinical, radiologic, and molecular analyses was conducted on a relatively large cohort of HG UTUC composed of 48 noninvasive and 118 invasive cases. If proven HG, current treatment of choice for UTUC was radical nephroureterectomy. However, no single cancer-specific death in patients with HG noninvasive UTUC was reported. Our findings suggest that despite an HG tumor, patients with noninvasive UTUC has excellent prognosis and may be overtreated by radical nephroureterectomy. Therefore, we aimed to find the clinical, radiologic, and pathologic characteristics that can aid to preoperatively predict the invasiveness of UTUC. For noninvasive UTUC, the following are helpful predictors: pattern-based pre-abdominal CT grading, absence of urothelial carcinoma in situ, presence of histologic intratumoral tumor grade heterogeneity, and preoperative cytologic findings of AUC or negative result (Fig. 6). These factors of clinical significance for predicting non-invasiveness were adopted to build a nomogram. According to the predictive value based on the risk assessment nomogram for non-invasiveness of HG UTUC, preoperative CT grading was the most powerful predictor (Fig. 6). Although not significantly correlated to tumor invasiveness, molecular analysis of the UTUC cases showed that HRAS mutation was significantly more frequent in cases with intratumoral tumor grade heterogeneity, while FGFR3 mutation inversely correlated with presence of urothelial carcinoma in situ and showed longer RFS.
Nomogram for predicting non-invasiveness in high grade (HG) upper tract urothelial carcinoma (UTUC). To estimate the probability of HG UTUC cases being noninvasive, each individual HG UTUC patients’ values were plotted on each variable axis. A vertical line shows the number of points that were assigned to the variable, and the points from each variable can be summed and projected to the bottom axis. According to the nomogram, preoperative computed tomography (CT) grading is the strongest predictive factor for noninvasive UTUC which is assigned with the highest points, followed by the presence of urothelial carcinoma in situ, the absence of discrepant urine cytology results, and absence of tumor grade heterogeneity.
In terms of preoperative staging of UTUC, CT may provide limited information whether tumor is T3 or higher or T2 or lower stage [6], given by the findings of peritumoral fat infiltration of T3 category. However, presence of peritumoral infiltration on CT is interpreted with high variability among expert and non-specialized radiologists. In this study, the CT criteria proposed by Honda et al. [6] was used with minute subdivision of the CT findings for UTUC staging. Considering the CT criteria, we were able to distinguish noninvasive UTUC from the invasive UTUC in the present study. However, there remains a risk of false negatives. Therefore, to improve preoperative diagnosis, various pathological and molecular biomarkers should be combined with CT findings to distinguish between invasive and noninvasive UTUC.
Next, noninvasive UTUC cases showed significant correlation to the presence of histologic intratumoral tumor grade heterogeneity (p=0.040), which possibly led to discrepant preoperative cytology results (p=0.018). However, intratumoral tumor grade heterogeneity itself did not have any impact on the survival rate (S8 Fig.). Tumor with intratumoral tumor grade heterogeneity might represent the “progressive” type, one of the two types of urothelial carcinogenesis pathways. Urothelial carcinogenesis, mostly studied from the urinary bladder, has two pathways: the first is the “progressive” type in which LG tumors with typical mutations of FGFR3 or HRAS acquire additional genetic alterations (TP53 or RB1) and/or obtain epigenetic changes resulting in HG tumors and the second, the “de novo” type that are HG tumors harboring TP53 and/or RB1 mutations and initially shows HG histology (urothelial carcinoma in situ) [18-20]. Prognostic difference between these two pathways is not known, and there is no validated genomic predictor for discovering patients at risk of LG-to-HG disease progression especially in UTUC. Similar to the previous study of Kittler et al. [21], we aimed to elucidate the underlying molecular mechanism of true tumor progression and de novo growth of UTUC by comparing pure HG UTUC and HG UTUC with intratumoral tumor grade heterogeneity. A phylogenic tree of the noninvasive UTUC cases that underwent WES showed shared HRAS and FGFR3 alteration throughout the spatially distant LG and HG components. There was no RB1 or P53 alteration in the private mutations of the HG components, which were consistent to the findings of Sfakianos et al. [7] in which TP53 or RB1 mutations were not the HG transforming factor in UTUC. However, only four cases underwent WES in the present study. Thus, other potential driver mutations responsible for HG transformation of UTUC should be further studied.
HRAS alteration was more prevalent in the renal pelvis (p=0.012), which may be associated with the higher pT category (p=0.034) and higher AJCC stage (p=0.045), since tumor in the renal pelvis may be diagnosed at an advanced stage compared with a tumor in the ureter, which easily causes symptom due to the narrow lumen. FGFR3 alteration was correlated to the absence of urothelial carcinoma in situ (p < 0.001), which may have led to its correlation to longer RFS. However, HRAS alteration did not have any survival impact. TCGA data was analyzed for comparison, but only five previous studies had available molecular results and limited information on tumor grade and stage [7,22-24]. Nevertheless, similar to our data, the TCGA data of the differentially expressed gene alterations between the invasive and noninvasive UTUC cases included FGFR3 alteration (32.1% vs. 63.6%). Similar to our findings, HRAS alteration was not directly associated with the invasiveness of UTUC. However, we believe there could be some selection bias in the TCGA analysis since only cases with grade and stage information were selectively used for our analysis. We believe the significance of HRAS alteration should be studied in a larger cohort in the future. Our findings can be supported by the recent study of Fujii et al. [9] that proposed five molecular subtypes of UTUC: hypermutated, TP53/MDM2, RAS (HRAS/KRAS/NRAS), FGFR3, and triple-negative. In this study, RAS-mutated tumors were all located in the renal pelvis, frequently showed HG histology, and had more aggressive phenotype than the FGFR3-mutated subtype and better prognosis than the TP53/MDM2-mutated and the triple-negative subtype [9]. Therefore, although the present study was not able to determine a statistical significance on survival rates, future studies should investigate the role of HRAS alteration in UTUC. Similar to our study, Sfakianos et al. [7] also performed a microdissection of HG and LG components of two UTUC cases and showed retained FGFR3 alteration. However, this study performed a targeted sequencing analysis of 10 genes, which did not include HRAS, and the selected cases had little information on the tumor stage or intratumoral heterogeneity [7]. Moreover, in the present study, HG UTUC cases showed a relatively high frequency of HRAS and FGFR3 mutation (noninvasive cases: 22.9% and 20.0%; invasive cases: 39.8% and 12.0%, respectively) compared to the reported rate of urinary bladder, although only hotstpot mutations for HRAS and FGFR3 mutations were analyzed. However, this finding is consistent with other studies that reported a higher prevalence of FGFR3, HRAS, and CDKN2B alteration (35.6% vs. 21.6%, p=0.065; 13.6% vs. 1.0%, p=0.001; and 15.3% vs. 3.9%, p=0.016, respectively), less alteration of TP53 and ARID1A (57.8% vs. 25.4%, p < 0.001 and 27.5% vs. 13.6%, p=0.050, respectively), and no RB1 mutation (0% vs. 18.6%, p < 0.001) in patients with UTUC than in those with urinary bladder [7]. Due to the relative rarity of UTUC, treatment guidelines such as the recommendation for the use of neoadjuvant chemotherapy are mostly based on urothelial carcinoma of the bladder [24]. Since UTUC harbors more targetable mutations such as HRAS and FGFR3 alteration even in the HG tumors, understanding the pathophysiology of UTUC would help in developing different treatment strategies.
With regard to CSS, our results from multivariate analysis showed that older age (p=0.044), higher pT category (p=0.019), higher AJCC stage (p=0.017), presence of LVI (p=0.031), and involvement of resection margin (p=0.013) are independent poor prognostic factors. Similarly, multivariate analysis of the RFS showed higher pT category (p=0.037), lymph node metastasis (p=0.005), preoperative distant metastasis (p=0.001), higher AJCC stage (p=0.025), presence of LVI (p < 0.001), PNI (p < 0.001), and urothelial carcinoma in situ (p=0.025) are independent prognostic factors. These findings including urothelial carcinoma in situ as a prognostic factor for recurrence are consistent findings with previous studies [11,25,26].
Our study has its strength in that a relatively large cohort of HG UTUC was comprehensively analyzed both radiologically and pathologically. Clinicopathologic factors including the grade and stage were accurately evaluated through a central review. However, the present study also has some limitations. First, the retrospective nature of the study might have caused a selection bias. Second, the number of cases for WES was too small to identify any driver mutations responsible for HG transformation, which require further study. Third, although all patients underwent gross examination with sampling of all areas of tumor with different appearance with at least one representative selection per 1 cm of the tumor, those with intratumoral tumor grade heterogeneity might have been missed due to the retrospective nature of the study. Lastly, the definition of “intratumoral heterogeneity” with a cutoff of 10% of the LG component may be somewhat ambiguous. Unfortunately, there is no standardized cutoff value for defining “intratumoral tumor grade heterogeneity”, however, since there is a generally accepted approach of designating HG urothelial carcinoma for those with HG component of ≥ 5% of the tumor, we have assessed our HG UTUC cohort for the amount of LG components by a 5% increment initially. As a result, most of the cases that showed areas of typical LG component showed easily identifiable LG components of ≥ 10% of the tumor, and no cases were considered with uncertainty between the two pathologists. Therefore, we believe this cutoff can be readily used in the clinical practice of pathologists. Future studies that will digitally quantify the HG and LG components can further strengthen our results.
In conclusion, although current treatment of choice for HG UTUC is radical nephroureterectomy, patients with HG noninvasive UTUC have excellent prognosis and have distinct clinical, radiologic, histologic, and molecular features in comparison to those with HG invasive UTUC. Thus, patients with HG noninvasive UTUC should be treated less aggressively. Our findings and predictive model for non-invasiveness in HG UTUC can be especially helpful in the prediction of tumor biology and prognosis for patients who are with risk factors such as old age, bilateral UTUC, or single kidney.
Electronic Supplementary Material
Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).
Notes
Ethical Statement
This study was approved by the Institutional Review Board of the Asan Medical Center (IRB No. 2021-0739), and patient’s consent was obtained.
Author Contributions
Conceived and designed the analysis: Ahn B, Cho YM, Kim D (Deokhoon Kim).
Collected the data: Ahn B, Hong B, Cho YM.
Contributed data or analysis tools: Ahn B, Kim D (Doeun Kim), Park KJ, Hong B.
Performed the analysis: Ahn B, Park JM, Yoon SY, Cho YM.
Wrote the paper: Ahn B, Kim D (Doeun Kim), Kim D (Deokhoon Kim), Park KJ, Cho YM.
Conflicts of Interest
Conflict of interest relevant to this article was not reported.
Acknowledgements
This study was supported by the Ministry of Science, ICT and Future Planning (2019R1A2C1088246) and a grant (2022IL00141) from the Asan Institute for Life Sciences, Asan Medical Center, Seoul, Republic of Korea.