Identification of Let-7f-5p as a novel biomarker of recurrence in non-muscle invasive bladder cancer

Abstract

BACKGROUND:

Among patients diagnosed with non-muscle invasive bladder cancer (NMIBC), 30% to 70% experience recurrences within 6 to 12 years of diagnosis. The need to screen for these events every 3 to 6 months and ultimately annually by cystoscopy makes bladder cancer one of the most expensive malignancies to manage.

OBJECTIVE:

The purpose of this study was to identify reproducible prognostic microRNAs in resected non-muscle invasive bladder tumor tissue that are predictive of the recurrent tumor phenotype as potential biomarkers and molecular therapeutic targets.

METHODS:

Two independent cohorts of NMIBC patients were analyzed using a biomarker discovery and validation approach, respectively.

RESULTS:

miRNA Let-7f-5p showed the strongest association with recurrence across both cohorts. Let-7f-5p levels in urine and plasma were both found to be significantly correlated with levels in tumor tissue. We assessed the therapeutic potential of targeting Lin28, a negative regulator of Let-7f-5p, with small-molecule inhibitor C1632. Lin28 inhibition significantly increased levels of Let-7f-5p expression and led to significant inhibition of viability and migration of HTB-2 cells.

CONCLUSIONS:

We have identified Let-7f-5p as a miRNA biomarker of recurrence in NMIBC tumors. We further demonstrate that targeting Lin28, a negative regulator of Let-7f-5p, represents a novel potential therapeutic opportunity in NMIBC.

1. Introduction

Currently, the U.S. has an estimated 500,000 patients with a history of urothelial carcinoma [1]. The high rate of disease recurrence and progression is a major challenge in patient management [2]. Of patients diagnosed with non-muscle invasive bladder cancer (NMIBC), 30–70% experience recurrence within 6–12 years of diagnosis, and 2–30% of tumors progress to muscle-invasive disease [3]. The need to screen for these events (every 3–6 months and ultimately annually by the invasive cystoscopy procedure) makes bladder cancer one of the most expensive malignancies, costing the U.S. an estimated $3.7 billion in 2001 [4, 5].

Recurrence rates vary considerably [6], even within a single histopathologic group [7]. Primary tumor clinicopathologic characteristics predictive of recurrence include multiplicity, tumor size, T category (depth of invasion), presence of carcinoma in situ, tumor grade, and patient gender [8]. Patients with low stage, low grade (LG) tumors can remain disease free for many years, but poorly differentiated tumors with a high grade (HG) often recur within one year and frequently progress to muscle invasive disease [2]. Management of these patients with TaHG, CIS or T1 tumors is clinically challenging. Bladder removal by cystectomy is performed in the subset of patients experiencing recurrent or progressing tumors that are refractory to treatment [9]. Early identification of those NMIBC patients likely to have progressing and refractory phenotypes would enable individualized treatment and surveillance recommendations, reducing patient burden and disease mortality.

Non-coding RNAs, particularly the microRNAs (miRNAs) have emerged as useful prognostic biomarkers in cancer in part because their small size makes them stable to degradation and thus robust to variations in sample handling [10]. The miRNAs regulate their target genes by binding to specific sites, usually in the 3’ untranslated region (UTR) of the target gene. The miRNA then modifies the target gene via translational repression, cleavage, degradation, or sequestration [11]. The objective of this project was to identify reproducible prognostic miRNAs in resected NMIBC tissue that are predictive of the recurrent tumor phenotype as potential biomarkers and molecular therapeutic targets.

2. Methods

All study procedures were approved by the Committee for the Protection of Human Subjects at Dartmouth College, Federal Wide Assurance (FWA#3095).

2.1 Patients

We utilized patients diagnosed with primary NMIBC in two independent cohorts for a biomarker discovery/ validation approach.

The New Hampshire (NH) Population-based Cohort was comprised of bladder cancer patients diagnosed in the state of NH between 2002–2004. Eligible cases were residents of the State of NH at the time of diagnosis identified using the State Cancer Registry, hospital pathology departments, and hospital cancer registries, as described. Information on bladder tumor clinicopathological features and recurrences was obtained from medical records, or provided by the treating hospital(s) (both in and outpatient records, including any pathology reports) covering the 2-year follow-up period ( $n=$ 169). We used an entire 10 $\mu$ M section for deparaffinization and RNA isolation using the Qiagen AllPrep FFPE tissue kit. Notably, we used entire tissue sections for this cohort with a large proportion of Ta low-grade tumors [12].

For the Dartmouth-Hitchcock Medical Center (DHMC) Cohort, we retrospectively selected a sequential set of bladder cancer patients identified through the hospital tumor registry diagnosed in the years 2008–2014. For this miRNA expression assessment project, we identified a subset of those patients with the clinically challenging non-muscle invasive histologic types: primary Ta High-Grade and T1 tumors (TaHG/T1) who had archived formalin-fixed, paraffin-embedded (FFPE) tissue blocks. We ensured that all tissue samples utilized represent the tissue that was removed prior to the administration of any intravesical immunotherapy or chemotherapy. We reviewed the patient medical records to ascertain clinical information related to recurrence and progression events. The study pathologists reviewed the hematoxylin and eosin (H&E)-stained slide and circled the non-cauterized tumor portion. For each tumor, we performed macrodissection on several of the matching 10 micron unstained tissue sections to select the circled portions. RNA was isolated from this portion using Qiagen Deparaffinization Reagent followed by the Qiagen AllPrep FFPE tissue kit (Qiagen Inc, Germantown, MD, USA) ( $n=$ 38).

2.2 Tissue-matched bio-fluid samples

On a subset of DHMC bladder cancer patients with tumor tissue samples, we also collected blood and urine samples post-diagnosis ( $n=$ 11 and $n=$ 16 samples, respectively) [13]. Whole blood was collected in a Vacutainer EDTA (K2) plastic tube (Becton Dickinson, CA, USA) and fractionated by centrifugation at 1,500 $\times$ g for 20 min at 20 ${}^{\circ }$ C. Plasma, enriched-white-blood cells, and red blood cells were aliquoted into cryogenic vials (Corning Incorporated, Corning, NY, USA) and stored at $-$ 80 ${}^{\circ }$ C. Urine was collected mid-stream in Clikseal containers (Therapak Corp, Buford, GA, USA), and subsequently transferred to 15-mL conical tubes (VWR, Radnor, PA, USA) for centrifugation. Urine was centrifuged 200 $\times$ g for 20 minutes, followed by 1000 $\times$ g 20 minutes to eliminate cell/debris to generate a cell-free urine sample. Final supernatant was aliquoted into cryogenic vials and stored at $-$ 80 ${}^{\circ }$ C.

Total RNA was extracted from (3) 20- $\mu$ M slices of FFPE tumor tissue using Norgen FFPE RNA/DNA Purification Plus Kit (Norgen Biotek Corp., Thorold, ON, Canada). Circulating and exosomal plasma microRNA was isolated from 200 $\mu$ L of plasma using Norgen Plasma/Serum Circulating and Exosomal RNA Isolation Kit. Cell-free miRNA from urine was obtained using Norgen Urine Exosome RNA Isolation Kit from 4 mL of cell-free urine. Enriched-WBC RNA was isolated using Norgen Total RNA Purification Kit. All protocols were performed according to manufacturer’s instructions. Plasma and urine miRNA was purified and concentrated using Amicon Ultra 0.5 columns (Millipore, Billerica, MA, USA), as described [14].

2.3 miRNA expression levels

The Dartmouth Genomics Shared Resource used the Nanostring human v3 microRNA expression assay (NanoString Technologies, Seattle, WA, USA) to quantify the levels of 800 miRNAs. Specific tags are ligated to the 3’ end of each miRNA. miRNA is then hybridized to a panel of miRNA:tagspecific nCounter capture and barcoded reporter probes. The nCounter Digital Analyzer counted individual fluorescent barcodes and quantified the target RNA molecules present in each sample. We used Nanostring Nsolver 3.0/4.0 software to normalize the count data to the positive controls and to average geometric mean of the top 100 detected miRNAs. We estimated the background level using the counts in the negative controls (mean $+$ 2SD), and restricted our analyses to the miRNAs expressed at counts above background in at least 1/3 of the tumors (330 miRNAs in the NH cohort; 169 miRNAs in the DHMC cohort).

2.4 Cell lines and reagents

The NMIBC cell line HTB-2 (RT4) was obtained from ATCC (Manassas, VA, USA). Cells were cultured in McCoy’s growth medium (Sigma-Aldrich, Burlington, MA, USA) with 10% fetal bovine serum and 1% Penicillin-Streptomycin.

2.5 RT-qPCR assays

HTB-2 cells (10 ${}^5$ cells/well) were seeded into flat-bottom cell culture 6-well plates and treated with dose ranges of Lin28 inhibitor C1632 (Tocris, R&D Systems, Inc., Minneapolis, MN, USA) for 48 h. RNA was extracted using the RNeasy Plus kit (Qiagen, Valencia, CA, USA) and treated with DNaseI. For RT-qPCR assays of Lin28A, Lin28B, and Actin, cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA), and RT-qPCR was performed using iQ SYBR Green Supermix (Bio-Rad). For RT-qPCR assays for Let-7f-5p miRNA expression, cDNA was synthesized using the TaqMan Advanced miRNA cDNA Synthesis Kit (Life Technologies Co., Thermo, Carlsbad, CA, USA), and cDNA amplification and RT-qPCR were performed using the TaqMan Advanced miRNA Assays (Life Technologies). The following primers were used: Lin28A (#Hs00702808_s1; Life Technologies), Lin28B (#Hs01013729_m1; Life Technologies), ACTB (F: TGACAGGATGCAGAAG GAGAT, R: GCGCTCAGGAGGAGCAAT), and Let-7f-5p (#478578_mir; Life Technologies).

2.6 Cell growth assays

HTB-2 cells were seeded in triplicate at 8,000/well in 96-well plates. The next day, cells were treated with dose ranges of C1632 for 5 d. Relative numbers of adherent cells were determined by sulforhodamine B (SRB) staining as described [15], and staining intensity was quantified by spectrophotometry (Abs ${}_{490}$ ).

Table 1

Non-muscle invasive bladder cancer patient characteristics by cohort

	NH population cohort		DHMC TaHG/T1 cohort
	$n=$ 169	%	$n=$ 38	%
Age
Mean $\pm$ SD	62.67 $\pm$ 10.36		69.84 $\pm$ 11.96
Gender
Female	47	28%	9	24%
Male	122	72%	29	76%
Multiplicity
Multi	50	30%	10	26%
Single	119	70%	28	74%
Size
$⩾$ 3 cm	69	58%	17	45%
$<$ 3 cm	50	42%	21	55%
Stage/grade
TaLG	131	78%	0	0%
TaHG	12	7%	10	26%
T1	26	16%	28	74%
Recurrence	98 events	58%	20 events	53%
Median	14.4 months		14.9 months

Table 2

NH population cohort top-ranking miRNAs for recurrence

Recurrence-free survival		Univariate		Multivariable ${}^{*}$
Recurrence events 98 of 169		$p$ -value	Coefficient	$p$ -value	Coefficient	HR ${}^{*}$	(95%CI)
hsa-let-7f-5p	$⩾$ vs. $<$ median	0.0069	$-$ 0.56	0.015	$-$ 0.53	0.59	(0.38–0.90)
hsa-miR-27b-3p	$⩾$ vs. $<$ median	0.18	0.27	0.091	0.35	1.42	(0.95–2.15)
Covariates:
Age	Years	0.2	0.013	0.11	0.02	1.02	(1.0–1.04)
Sex	Female vs. male	0.69	1.09	0.54	0.14	1.16	(0.73–1.82)
Grade	High vs. low	0.18	0.33	0.32	0.32	1.39	(0.73–2.64)
Stage	Ta vs. T1	0.89	0.038	0.34	$-$ 0.34	0.71	(0.35–1.43)
Multiplicity	Multiple vs. single	0.0095	0.55	0.035	0.47	1.6	(0.98–1.0)

Among 331 detectable miRNAs. ${}^{*}$ Multivariable model adjusted for sex, age, multiplicity, stage, grade.

Table 3

Patient and tumor characteristics stratified by let-7f-5p

hsa-let-7f-5p	$<$ Median	$⩾$ Median	Univariate
	$n=$ 85	$n=$ 84	$p$ -value
Age (mean (SD))	61.18 (10.36)	64.18 (10.20)	0.059
Sex (%)
Male	58 (68.2)	64 (76.2)	0.33
Female	27 (31.8)	20 (23.8)
Grade (%)
Low	66 (77.6)	71 (84.5)	0.35
High	19 (22.4)	13 (15.5)
Stage (%)
Ta	70 (82.4)	73 (86.9)	0.54
T1	15 (17.6)	11 (13.1)
Multiplicity (%)
Single	53 (62.4)	66 (78.6)	0.032
Multiple	32 (37.6)	18 (21.4)

2.7 Migration/scratch assay

HTB-2 cells were plated in triplicate and grown to 90% confluency in flat bottom cell culture 12-well plates (Corning, NY, USA). Monolayers were scratched using a 1-mL pipette tip, and dead cells were washed off. Cells were grown for an addition 72 h, after which adherent cells were fixed and stained with 0.5% crystal violet in 20% methanol. Plates were scanned using an Epson Perfection V600, and migration distance was measured using the ruler function in Adobe Photoshop software.

2.8 Target gene analysis

To assess a priori target genes, we performed a secondary analysis of data from bladder tumors assayed as part of the National Cancer Institute’s Genomic Data Commons (GDC) project, restricting to $n=$ 134 patients in AJCC pathological Stage I or II. Expression levels of high-mobility group A2 (HMGA2) and the sodium iodide symporter (SLC5A5) were assessed by HTseq Fragments per Kilobase of transcript per Million mapped reads (FPKM), with upper quartile normalization, and Let-7f levels were assessed by miRNAseq [16].

2.9 Statistical analyses

We defined the first recurrent tumor as any tumor identified following a disease-free remission period, more than 90 d after the date of initial primary bladder tumor diagnosis. These recurrent tumors include subsequent tumors of the same level of invasiveness, as well as those progressing to a higher stage/grade. Persistent primary tumors that did not have a remission period were excluded from the analysis of recurrence. Time to recurrence was calculated as the time between the initial diagnosis date and the date of the first recurrence event. For progression, the event was the diagnosis of a tumor with a greater stage or grade than the initial primary bladder tumor. If no events were reported, the date the patient was last seen documented in the medical record was used for censoring.

Median times to first recurrence, progression or survival were calculated using the Kaplan-Meier method. Multivariable analysis of time to the first bladder tumor recurrence, progression and survival analyses were performed using Cox-proportional hazards regression analysis. We tested the Cox model proportional hazards assumption by checking the independence of the scaled Schoenfeld residuals with time, and that the deviance residuals were symmetric. The standard prognostic model included adjustment for age at diagnosis of first bladder tumor, gender, tumor size ( $<$ 3, 3 $+$ cm), multiplicity (single, multiple), stage (Ta, T1)/grade (low-grade, high grade) in the model. We used the Spearman test to assess the correlation between Let-7f-5p levels in urine or plasma and tumor tissue. One-way ANOVA assessed differences in expression after Lin28 inhibition in cell culture. $p$ values represent the results of two-sided statistical tests. Analyses were performed using R 3.4.1.

Figure 1.

Recurrence-free survival prognosis of NH and DHMC cohort by Let-7f-5p levels in tumor tissue. Kaplan-Meier plots depict the probability of recurrence stratified by let-7f-5p levels $⩾$ Median (black line) or $<$ Median (gray line) in tissue samples from the NH cohort (A) and the DHMC cohort (B). Patients in both the NH and DHMC cohorts with let-7f-5p levels above the median had longer time to recurrence than those with levels below the mean ( $p=$ 0.015 and 0.065, respectively).

Figure 2.

Correlation of Lef-7f-5p levels in urine and plasma versus tumor tissue. Tissue-matched bio-fluid specimens from the DHMC cohort were assayed for let-7f-5b levels relative to tumor tissue. Let-7f-5p expression levels in both urine ( $n=$ 16, $p=$ 0.004) (A) and plasma ( $n=$ 11, $p=$ 0.034) (B) are correlated with expression levels in tumor tissue by Spearman testing.

Figure 3.

The effect of Lin28 inhibition on Let-7f-5p expression, cell viability, and cell migration in non-muscle invasive bladder cancer cells. (A) Let-7f-5p levels in HTB-2 cells were analyzed by qPCR following treatment with Lin28 inhibitor C1632 for 48 hours at the indicated concentrations (x-axis, $\mu$ M in DMSO). Expression levels were normalized to the no treatment control. Experiment was performed in triplicate wells. (B) Cell viability of HTB-2 cells following treatment with C1632 was quantified by Sulforhodamine B (SRB) assay. Cell viability was normalized to the no treatment control. Experiment was performed in triplicate wells. (C) Migration of HTB-2 cells following treatment with C1632 was assessed by scratch assay. Distance was measured using Adobe Photoshop, and normalized to the baseline (0 hour) control. Error bars represent standard deviation. Experiment was performed in triplicate wells, with 4 distances measured in each well. ${}^{**}$ Benjamini and Hochberg FDR-adjusted $p<$ 0.0001, ${}^{*}p<$ 0.05 by 1-way ANOVA.

3. Results

As shown in Table 1, the NMIBC patient cohorts assessed for miRNAs included a majority of male patients (72–76%) diagnosed with bladder cancer in their 60’s. Multiple tumors were present in a subset of patients (26–30%), and less than half of patients had large tumors ( $>$ 3 cm 42–45%). The majority of the patients in the NH population cohort had Ta low-grade tumors (78%). In contrast, we restricted the DHMC cohort to patients with more clinically challenging non-muscle invasive tumors (26% Ta high-grade, 74% T1).

We used the NH population cohort as a ‘discovery’ set to identify tumor tissue miRNAs associated with time-to-first recurrence in a multivariable model adjusted for sex, age, multiplicity, tumor size, stage, and grade. miRNA expression was stratified by the median value. Out of the 331 detectable miRNAs ranked by p-value, Let-7f-5p had the strongest association with recurrence in the NH population cohort ( $p=$ 0.015, Table 2). The second strongest association with recurrence in the NH population cohort was miR-27b-3p ( $p=$ 0.091, Table 2). We then used the DHMC cohort to assess the markers in an independent high-risk tumor situation, and found that only Let-7f-5p ( $p=$ 0.065), but not miR-27b-3p ( $p=$ 0.63), was consistent in its association with recurrence. The predictive accuracy measures of Let-7f-5p at the median for recurrence at 12-months in the DHMC cohort were: sensitivity 32.29 $\pm$ 13.58, specificity 60.0 $\pm$ 11.1, PPV 28.89 $\pm$ 12.63, NPV 63.77 $\pm$ 10.58. Kaplan-Meier plots for recurrence-free survival patients show that both the NH and DHMC cohorts with Let-7f-5p above the median had longer time-to-recurrence (Fig. 1a and b). Among patients with levels of Let-7f-5p expression above the median, the recurrence Hazard Ratio (HR) in tumor tissue was 0.58 (0.38–0.88) (Table 2, $p=$ 0.015 adjusted for sex, age, multiplicity, stage, grade). Stratifying by Let-7f-5p at the median, we observe a lower proportion of patients with multiple tumors at baseline in patients with high Let-7f-5p levels ( $⩾$ median 21.4% multiple, $<$ median 37.5% multiple, chi-square test $p=$ 0.032) (Table 3).

To establish if blood and urine could also be used for minimally-invasive assessment of prognostic miRNA levels, we used a set of blood ( $n=$ 11) and urine samples ( $n=$ 17) that were patient matched to tumor tissue and assessed the levels of the tumor tissue-prognostic let-7f-5p in the corresponding bio-fluid samples. Levels of Let-7f-5p expression in tumor tissue were found to be correlated with Let-7f-5p expression levels in urine ( $p=$ 0.0142) and in plasma exosomes ( $p=$ 0.0004) (Fig. 2).

Because post-transcriptional downregulation of let-7 miRs can be attributed to Lin28 activation [17], we asked if inhibition of Lin28 with small molecule Lin28 inhibitor C1632 would result in increased let-7f-5p levels in NMIBC cell line HTB-2. Indeed, treatment of HTB-2 cells with Lin28i induced a dose-dependent increase in let-7f-5p (Benjamini Hochberg FDR-adjusted $p<$ 0.0001 by 1-way ANOVA), but not Lin28A, Lin28B, or Actin control (Fig. 3 and Supplemental Fig. 1). Furthermore, treatment with C1632 reduced cell viability and migration in HTB-2 cells in a dose response manner ( $p<$ 0.0001 for both; Fig. 3 and Supplemental Fig. 2).

We also used publically available data to perform a secondary analysis of Let-7f target genes in bladder tumor tissue, restricting to $n=$ 134 patients in AJCC pathological Stage I or II. Supplemental Fig. 3 shows a statistically significant inverse correlation between Let-7f and levels of the hypothesized target gene HMGA2 ( $p=$ 0.0051), but not of SLC5A5 ( $p=$ 0.83).

4. Discussion

In the present study, we sought to identify prognostic miRNAs in NMIBC tumor tissue. We used patient cohorts representing two different clinical settings: population-based cohort representing patients who were residents of New Hampshire, including those treated in community hospitals, and a referral hospital cohort representing patients seen at Dartmouth-Hitchcock Medical Center. We identified let-7f-5p as a prognostic miRNA that reliably corresponds with probability of recurrence-free survival in both cohorts examined. Furthermore, we found that let-7f-5p levels in tumor tissue correlate with levels in plasma and urine, which can be collected without biopsy. Thus, Let-7f-5p expression may have clinical utility as a biomarker of recurrence.

The let-7 family of miRNAs is comprised of thirteen different members found on nine different chromosomes, and expression of let7 miRNAs are frequently downregulated in multiple tumor types [18]. Importantly, let7 miRNAs are thought to regulate tumorigenesis by targeting multiple oncogenic factors, including c-MYC, Ras, high-mobility group A (HMGA), and others, by targeting transcripts of these proteins directly or by targeting their downstream effectors [19]. Manipulating Let-7f levels in cell lines from the thyroid and the lung modified levels of HMGA2, and of the sodium iodide symporter (NIS; SLC5A5) [20, 21, 22, 23, 24]. Our analysis of levels of these genes in bladder tumors showed Let-7f has a statistically significant inverse correlation with HMGA2, but not with SLC5A5.

Because of its action as a tumor suppressor miRNA, reduction of let-7 levels may enable the oncogenic activity of these factors, leading to tumor formation and progression. Due to the importance of let-7 miRNAs in tumorigenesis, these miRNAs may be useful biomarkers with prognostic value. Indeed, let-7f has been identified as an indicator of early stage ovarian cancer, and is also downregulated in the tumor tissue and plasma of colorectal carcinoma patients [25, 26]. To our knowledge, the prognostic utility of let-7 miRNAs assessed in urine and plasma biofluids in the context of NMIBC has not been explored. Urinary exosomal Let-7f levels were 1.73-fold lower in high-grade bladder cancer patients ( $n=$ 18) compared to healthy controls ( $n=$ 9) $p=$ 0.096 [27]. While there has been some evidence of other miRNA biomarkers of recurrence in NMIBC [12], these markers have yet to achieve utility in the clinic [28]. Our data highlights let-7f-5p as a strong candidate miRNA biomarker that can be evaluated in non-invasive specimens such as urine and in tumor tissue may have prognostic value in NMIBC. A logical next step would be to determine whether let-7f-5p can be combined in tandem with other characterized biomarkers of recurrence such as TERT and FGFR3 mutations in urinary cell-free DNA [29, 30].

There is currently a critical need for new therapeutic strategies to prevent recurrence in NMIBC. Our findings suggest that Lin28, a negative regulator of let-7 miRNAs, may be a therapeutic target for NMIBC. Lin28 is an RNA binding protein whose oncogenic activity is attributed to its role in selectively targeting let-7 for degradation by DIS3L2 exonuclease [31]. Inhibition of Lin28 by the small molecule C1632 (Lin28i) can inhibit stemness and induce differentiation of mouse embryonic stem cells, as well as inhibit proliferation of human cancer cells [32]. Accordingly, we treated NMIBC cell line HTB-2 with Lin28i and observed an increase in let-7f-5p, suggesting Lin28 acts as a negative regulator of let-7f-5p in these cells (Fig. 3). Further analysis showed that Lin28i reduced HTB-2 cell viability, colony formation, and migration (Supplemental Figure). Taken together, these observations suggest that Lin28 inhibition may reactivate the tumor-suppressive functions of let-7f-5p and could serve as a novel therapeutic opportunity in NMIBC.

A limitation of our study is that the cohorts may not be generalizable to populations with other demographics. New Hampshire has proportionally higher rates of bladder cancer diagnosis compared to the rest of the country, found to be due at least in part to higher levels of arsenic in drinking water [33]. Additional national and international studies are thus required to generalize these findings to the broader population. An additional limitation is sample size, particularly in the DHMC cohort, which may have limited interpretations of significance in our validation analysis. However, despite these limitations, our study suggests that determination of let-7f-5p expression may highlight patients at elevated risk for recurrence, that let-7f-5p represents an optimal biomarker because expression in tumor tissue correlates with expression inin urine and plasma, and that targeting the Lin28-Let-7 axis may represent a novel therapeutic opportunity in NMIBC. These findings offer potential precision medicine opportunities with important clinical implications, and warrant further study.