Improving multiparametric MR-transrectal ultrasound guided fusion prostate biopsies with hyperpolarized ¹³C pyruvate metabolic imaging: A technical development study

Abstract

Purpose:

To develop techniques and establish a workflow using hyperpolarized carbon-13 (¹³C) MRI and the pyruvate-to-lactate conversion rate (k_PL) biomarker to guide MR-transrectal ultrasound fusion prostate biopsies.

Methods:

The integrated multiparametric MRI (mpMRI) exam consisted of a 1-min hyperpolarized ¹³C-pyruvate EPI acquisition added to a conventional prostate mpMRI exam. Maps of k_PL values were calculated, uploaded to a picture archiving and communication system and targeting platform, and displayed as color overlays on T₂-weighted anatomic images. Abdominal radiologists identified ¹³C research biopsy targets based on the general recommendation of focal lesions with k_PL >0.02(s⁻¹), and created a targeting report for each study. Urologists conducted transrectal ultrasound-guided MR fusion biopsies, including the standard ¹H–mpMRI targets as well as 12–14 core systematic biopsies informed by the research ¹³C-k_PL targets. All biopsy results were included in the final pathology report and calculated toward clinical risk.

Results:

This study demonstrated the safety and technical feasibility of integrating hyperpolarized ¹³C metabolic targeting into routine ¹H–mpMRI and transrectal ultrasound fusion biopsy workflows, evaluated via 5 men (median age 71 years, prostate-specific antigen 8.4 ng/mL, Cancer of the Prostate Risk Assessment score 2) on active surveillance undergoing integrated scan and subsequent biopsies. No adverse event was reported. Median turnaround time was less than 3 days from scan to ¹³C-k_PL targets, and scan-to-biopsy time was 2 weeks. Median number of ¹³C targets was 1 (range: 1–2) per patient, measuring 1.0 cm (range: 0.6–1.9) in diameter, with a median kPL of 0.0319 s⁻¹ (range: 0.0198–0.0410).

Conclusions:

This proof-of-concept work demonstrated the safety and feasibility of integrating hyperpolarized 13C MR biomarkers to the standard mpMRI workflow to guide MR-transrectal ultrasound fusion biopsies.

1 |. INTRODUCTION

Multiparametric prostate MRI (mpMRI) is a standard-of-care imaging tool in the clinical workup of men with either known or suspected prostate cancer. Many practice guidelines such as National Comprehensive Cancer Network, American Urological Association, and European Association of Urology¹ have incorporated mpMRI into the workflow of prostate cancer diagnosis, although the indications for which mpMRI should be ordered, and at what frequency, still vary from guideline to guideline. This stems from divergent views regarding mpMRI’s overall role and importance in the prostate cancer risk assessment.

Active surveillance is a preferred management strategy for men with low-risk, and selected intermediate-risk, localized prostate cancer to minimize treatment-associated morbidity without compromising oncologic outcomes. This is evidenced by the 10-year cancer-specific survival rate of nearly 99% for men undergoing active surveillance.² Whereas conventional mpMRI is widely used to guide prostate biopsies, its role in the surveillance setting has long been a subject of controversy.^3,4 Although mpMRI-guided confirmatory biopsy led to decreased active surveillance failures,⁵ a few randomized and retrospective studies failed to confirm the clinical utility of mpMRI in active surveillance.^6–8 These contradicting results not only lead to discrepant endorsement among guidelines regarding the use of surveillance mpMRI but highlighted the unmet need to improve both the diagnostic accuracy and yield of mpMRI in this setting.

Hyperpolarized (HP) carbon-13 (¹³C)-pyruvate MRI is a new rapid molecular imaging technique that can detect increased pyruvate-to-lactate conversion rates (k_PL) in clinically significant, aggressive prostate cancer as compared to more indolent tumors.^9,10 The emerging technology uses dynamic nuclear polarization to increase the SNR of ¹³C-enriched compounds by more than 50,000-fold through means of temporarily rearranging the spins to increase their nuclear polarization.^11,12 This enables interrogation of previously inaccessible in vivo metabolic pathways with unprecedented signal quality and quantitative accuracy. Several studies have explored the technical aspects of HP ¹³C MRI in men with localized and metastatic prostate cancer,^13–15 as well as correlating 13C markers to histological and molecular signatures of this malignancy.^16,17

MR-guided transrectal ultrasound (TRUS) fusion biopsy was designed to utilize MRI’s superior tissue contrast compared to ultrasound to detect prostatic lesions and improve prostate cancer diagnosis and grading. The fusion technology combines targeting information from a prior diagnostic MR exam with real-time, intraprocedural ultrasound guidance for accurate prostate tissue sampling.¹⁸ MR-guided TRUS fusion biopsies, together with systematic biopsies, are usually conducted in urologists’ offices as a simple outpatient procedure without requiring an operating room or general anesthesia.

Thus far, HP ¹³C MRI of localized prostate cancer has been studied primarily in the context of a high-risk cohort, looking at pathological correlation with the prostatectomy specimen.^16,17 Prior to this study, it had not been applied to prospectively guide confirmatory or surveillance biopsies. This technical development project aimed to integrate the metabolically defined research targets based on HP ¹³C-pyruvate MRI into mpMRI workflow and guide MR-TRUS fusion biopsies to improve the identification of clinically significant disease.

2 |. METHODS

2.1 |. Hyperpolarized ¹³C MRI and standard-of-care multiparametric MRI protocols

The MRI exams were conducted on a clinical 3 Tesla MRI scanner (MR750, GE Healthcare, Chicago IL) equipped with multinuclear spectroscopy capability. The MR coils and imaging setup have been previously described.¹⁹ Briefly, HP ¹³C imaging was performed using a clamshell Helmholtz transmitter and an endorectal coil for receive. For proton imaging, the body coil was used as a transmitter, and signal reception was accomplished using a 4-channel torso array combined with the endorectal probe. The endorectal receiver was a specialized proton (¹H)/¹³C dual-element design.¹⁹

The proton mpMRI exam consisted of T₁-weighted fast spin echo (FSE), T₂-weighted FSE (in-plane resolution = 0.35 mm, 3 mm slices, TR/TE = 6000/102 ms), 3D T₂-FSE, and small FOV (Field-of-view optimized and constrained undistorted single-shot, FOCUS) DWI (TR/TE = 4000/78, pixel bandwidth = 1305, b = 0 and 600 s/mm², and 3 mm slice thickness) sequences. Dynamic contrast-enhanced imaging using a 3D fast spoiled gradient-recalled echo (SPGR) sequence (TR/TE = 3.5/0.9 ms, flip angle = 5°, and 3 mm slice thickness) with gadobutrol (Bayer Pharmaceuticals, Leverkusen, Germany) was acquired as the final series of the exam after the conclusion of the ¹³C-pyruvate acquisition and other proton imaging. The multiparametric portion of the exam was consistent with the standard mpMRI²⁰ indicated for prostate cancer care at the University of California, San Francisco (UCSF).

For the hyperpolarized ¹³C pyruvate study, pharmacy kits containing a mixture of 1.432 g of GMP grade [1-¹³C] pyruvic acid (MilliporeSigma, Miamisburg OH) and 28 mg electron paramagnetic agent (AH111501; GE Healthcare, Oslo, Norway) were prepared and polarized in a 5 Tesla clinical-research polarizer (SPINLab, GE Healthcare, Chicago IL) at 0.77 K for 2.5–3 hours. Subsequently, the pyruvic acid was rapidly dissoluted and neutralized with tris-EDTA buffer solution, yielding sterile doses meeting release criteria of pyruvate concentration (median: 248 mM, 238–271 mM), polarization (median: 38.4%, 34.6%–40.6%), pH (median: 7.7, 7.4–8.2), temperature (<37 °C), and electron paramagnetic agent concentration (median: 1.5 μM, 0.7–1.9 μM). Following terminal sterilization, pharmaceutical release was issued after confirming quality control parameters and filter bubble point test as previously published.²¹ The dosage of pyruvate solution delivered to the patient was 0.43 mL/kg at an injection rate of 5 mL/s (up to 40 mL), immediately followed by a 20 mL saline flush at the same rate.

A symmetric EPI sequence was prescribed for the HP ¹³C study.²² Key sequence parameters included TR/TE = 1 s/25.2 ms; resolution = 6.5–8 mm in-plane, 8 mm through-plane; FOV = 10.4 × 10.4–12.8 × 12.8 cm in-plane, and 11.2 cm through-plane. Independent flip angles were applied to pyruvate (15°) and lactate (30°) resonances. The acquisition started 10 s after the completion of the pyruvate injection and saline flush. A rapid, low-resolution T₂-FSE axial series (in-plane resolution = 0.35 × 0.8 mm, 3 mm slices, ~1 min length) was acquired immediately after the ¹³C scan to measure possible motion shift during the exam, which could result from patient movements or bladder filling, and the resulting misalignment in between series. ¹³C center frequency and power were calibrated using a built-in 8 M urea phantom inside the endorectal coil. Transmit power was increased to ~133% of the nominally calibrated power to compensate for the inductive coupling losses between clamshell transmitter and the ¹³C element of the endorectal coil.

2.2 |. Image processing

The ¹³C-pyruvate EPI images were reconstructed using the GE Orchestra Toolbox (GE Healthcare, Chicago IL). Nyquist ghost correction was conducted using the ¹H reference scan method, as previously described.²² Global phases of the pyruvate and lactate images were independently calculated and accounted for. The HP ¹³C metabolic biomarker, k_PL, was quantified using an inputless 2-site exchange model that makes no assumptions about the dynamic or bolus profile of pyruvate but rather derives k_PL values solely based on measured pyruvate and lactate signals for improved robustness.²³

Any misalignment was manually calculated between the high-quality T₂-weighted series, acquired near the beginning of the exam, and the quick, post-¹³C-injection T₂-FSE series, and the ¹³C data were shifted accordingly in 3D to compensate for motion offset. This improved the alignment precision between the k_PL maps and the high-quality T₂ series on the targeting software Dynacad (Philips Invivo Corp., Gainsville, FL), and thereby allowed more accurate lesion identification and outlining.

2.3 |. Transfer of ¹³C-pyruvate MRI data to PACS and targeting software

The k_PL maps were interpolated to the T₂-FSE resolution, matching the exact same registration and matrix size as the T₂ series for easy visualization and overlays. A predefined series number was assigned to the k_PL maps. For example, if the high-quality T2 were series 5, the k_PL map would have been series 502.

The k_PL map series was labeled as nondiagnostic HP-13C kPL (diffusion) to easily distinguish it from clinical ¹H mpMRI sequences. The color k_PL map was overlaid on T₂-weighted images. Using the keyword “(Diffusion)” masqueraded the k_PL maps as a diffusion-weighted series. This nomenclature allowed us to repurpose the built-in color overlay function on Dynacad (Philips Invivo Corp.; referred to as “fusion” overlays on the UI) that was intended for overlay of DWI as a pseudo color map over grayscale T₂-weighted images.

The k_PL maps were first uploaded to the picture archiving and communication system (PACS). Following a quality control check on one of the PACS workstations to ensure correct registration, the k_PL maps were transferred to the Dynacad (Philips Invivo Corp.), a commercial targeting software that radiologists routinely use to identify and outline suspicious prostate MRI lesions as a part of biopsy planning. All the processing and data transfer occurred between our internal radiology network and PACS system and were therefore compliant with the Health Insurance Portability and Accountability Act.

2.4 |. Prostate lesion targeting and fusion biopsy

Figure 1 illustrates the workflow of HP ¹³C MR research targeting of the prostate. The conventional ¹H mpMRI exam was interpreted by board-certified radiologists according to the standard departmental workflow, and any lesions were categorized using Prostate Imaging Reporting and Data System (PIRADS) v2.1.²⁴ Any suspicious lesions were outlined in Dynacad (Philips Invivo Corp.) in preparation for targeted biopsy. One of 3 board-certified abdominal radiologists, each with more than 10 years of experience reading prostate MRI, additionally outlined the ¹³C research targets on the k_PL-T₂w color overlay displayed on Dynacad (Philips Invivo Corp.) (Figure 2A). The ¹³C research targets were identified and delineated following the general recommendation of delineating focal lesions on the k_PL map with k_PL value >0.02(s⁻¹), a biomarker of suspected cancer.¹⁷ The recommended k_PL threshold of 0.02(s⁻¹) for this initial technical development study was selected based on a prior high-risk cohort of patients (Supporting Information Figure S1), who after ¹³C-pyruvate MRI scans underwent radical prostatectomy with step-section histopathology (gold standard), representing the k_PL differences of high- versus low-grade tumors (corrected for different MR sequence TEs²⁵). Radiologists’ experience and judgment play an important role in target selection (similar to selection of ¹H MRI targets). Their discretion to include/exclude a target is allowed based on lesion focality, shape, and visual features, as well as likelihood of tumor presence in a given anatomical zone. The ¹H mpMRI-defined clinical targets and ¹³C-defined research targets were independently labeled for urologist review.

FIGURE 1.

The workflow developed in this project for HP ¹³C MR research targeting of prostate biopsies based on abnormally high pyruvate-to-lactate conversion k_PL values. The HP ¹³C MR exam and research targeting were integrated into the SOC MRI fusion and systematic biopsy procedures at our institution. First, the patient undergoes an integrated mpMRI exam of the prostate, including a 1-min acquisition following the HP ¹³C-pyruvate injection. The k_PL map is calculated and uploaded to PACS and a software targeting platform (Dynacad, Philips Invivo Corp., Gainsville FL). A radiologist reads the study and outlines the research targets based on ¹³C k_PL findings, in addition to those from the PIRADS lesions based on the ¹H mpMRI. The targets and a report are uploaded to the fusion biopsy system (UroNav, Philips Invivo Corp., Gainsville FL) in the urologist’s offices, where they review the targeting and plan for the procedure. After US/MRI fusion-guided biopsies are performed, the tissue specimens are submitted to Pathology for processing and diagnosis. Thus, the HP ¹³C research biopsy integration takes advantage of the existing infrastructure and minimizes the additional workload for the researchers and clinicians involved. ¹³C, carbon-13; HP, hyperpolarized; k_PL, pyruvate-to-lactate conversion rate; mp, multiparametric; PACS, picture archiving and communication system; PIRADS, Prostate Imaging Reporting and Data System; SOC, standard- of- care; US, ultrasound.

FIGURE 2.

(A) A representative targeting protocol using a commercial prostate biopsy targeting platform (Dynacad, Philips Invivo Corps.). This protocol can also be found in Supporting Information Video S1. The 3D k_PL image series was named with the keyword “diffusion” to allow a fusion overlay, displaying k_PL as a pseudocolor over T₂-weighted series. The overlay was displayed side by side with T₂, ADC, and DCE maps, enabling the radiologist to correlate between series and outline 3D ROI for both ¹H PIRADS and ¹³C research biopsy targets. Whereas the recommended k_PL threshold for identifying potentially high-grade ¹³C lesions was set to 0.02(s⁻¹), the lowest value of the heatmap was set to 0.01 for display purposes. This is designed to provide radiologists context on the shape/size of the lesion. The corresponding k_PL scales is shown next to the original color bar. (B) Both clinical and research targets are transferred to a commercial TRUS-fusion biopsy platform (UroNav, Philips Invivo Corps.), where the research biopsy targets were counted as systematic biopsies. The urologist sampled these targets under TRUS fusion guidance during a biopsy session, assisted by TRUS-MRI fusion (left panel: US axial, top right: MR sagittal) and 3D-rendered segmentation (bottom right panel) of the prostate. The biopsied tissues were submitted for histopathology analyses. ¹H, proton; ROI, region of interest; TRUS, transrectal ultrasound.

Both ¹³C research and standard ¹H mpMRI targeted biopsies were conducted by 1 of the 3 board-certified urological oncologists, each with more than 10 years of experience, who used a commercial fusion-biopsy platform (Uronav, Philips Invivo Corp., Gainsville, FL) following the standard mpMRI-targeted biopsy procedure at our institution (Figure 2B). The transrectal biopsy approach was used by our urologists per institutional practice. The MRI-generated overlays delineating ¹³C/¹H targets were fused with real-time TRUS images in the UroNav system (Philips Invivo Corp.) to guide the biopsy sampling. Systematic TRUS biopsies (12–14 cores) were also conducted in the same session, and the biopsy cores of any ¹³C research target replaced the systematic biopsy in the same sextant. As such, the ¹³C research biopsy did not increase the total number of cores.

A sample overlay and targeting procedure is illustrated in Supporting Information Video S1.

2.5 |. Patient characteristics

Five patients with biopsy-confirmed prostate cancer were enrolled (NCT03933670). Eligible patients were 18 years or older, had a biopsy-confirmed diagnosis of prostate cancer, ECOG score 0 or 1, and were either candidates for or currently on an active surveillance protocol as defined by the UCSF urologic oncology practices at the time of enrollment.²⁶ The key exclusion criteria entailed prior treatments for prostate cancer, biopsy within 14 days prior to ¹³C MRI, poorly controlled hypertension, or contraindication for endorectal coil placement. The patient studies were conducted with informed consent in compliance with Food and Drug Administration- and Institutional Review Board-approved protocols (NCT03933670).

2.6 |. Pathology assessment

The specimens from the ¹³C research biopsies were submitted to pathology along with the other samples from the same session. All formalin-fixed–paraffin-embedded cores were read by experienced urologic pathologists in a standardized fashion²⁷ and included in the final pathology report. UCSF–Cancer of the Prostate Risk Assessment (CAPRA) score²⁸ was recalculated based on the updated biopsy findings and the most recent prostate-specific antigen values.

3 |. RESULTS

3.1 |. Safety and technical feasibility

Integrated MRI exams and the ensuing biopsies for all 5 patients were safe, successful, and without adverse events. The k_PL (HP ¹³C metabolic biomarker) map was calculated and uploaded to PACS/Dynacad (Philips Invivo Corp.) along with the mpMRI exam. Image postprocessing and upload of k_PL maps were done the same day, typically within 1–2 h after the end of the exam. The average turnaround time for the MRI report and targeting was less than 3 days after each exam. A ¹³C research biopsy targeting report (a representative instance given in Figure 3) was generated and showed the target locations in the 3D segmented prostate, as well as the ¹³C-k_PL/T₂ overlay, diffusions, and T₁-weighted images arranged side by side to assist the urologists planning the biopsies.

FIGURE 3.

Shows a representative biopsy targeting report a radiologist created using DynaCAD (Philips Invivo Corp.). The report and targets were then sent to the UroNav system (Philips Invivo Corp.) to assist the urological oncologist to identify the ¹H mpMRI (PIRADs) and ¹³C research targets and plan for the biopsy procedure. The report, shown as montage here, illustrates the target locations on the 3D segmented prostate for visual reference (left panel). In the center panel, the ¹³C-k_PL/T₂ overlay, DWI, and T₁-weighted images are arranged side by side. A ¹³C-k_PL lesion was identified and outlined at the right mid-PZ. The right panel reports the automatically calculated volumes and mean ADC values over the outlined ¹³C target. PZ, peripheral zone.

3.2 |. HP ¹³C MRI targeting and MR-guided TRUS fusion biopsy

The patient demographics (N = 5) and clinical characteristics are summarized in Figure 4. The patients enrolled in this study had low- to intermediate-risk disease with median age 71 (range: 62–79), prostate-specific antigen 8.4 ng/mL (range: 1.3–17), and CAPRA score 2 (range: 1–3). The median number of ¹³C research targets was 1 (range: 1–2), and that of proton mpMRI was 1 (range: 0–2). The ¹³C targets measured 1 cm (range: 0.6–1.9) in diameter; the median k_PL was 0.0319 s⁻¹ (range: 0.0198–0.0410); and the intralesion distribution is given in Table 1. The mean k_PL in the segmented prostate excluding ¹³C targets was 0.0110±0.0022 s⁻¹. The index lesions on proton mpMRI had a median PIRADS score 4 (range: 2–5). The ¹H targets had a median 1.1 cm (range: 0.9–2) diameter. All 5 patients underwent TRUS/MRI fusion and systematic biopsies after the integrated mpMRI exam, with 2–3 cores sampled per target.

FIGURE 4.

(A) Pie chart summarizing the serum PSA and age of this initial cohort. (B) Pie chart summarizing the pathologic characteristics of the HP ¹³C research biopsies, PIRADS scores of ¹H mpMRI biopsies, overall Gleason score, and clinical risk (CAPRA score). The Gleason 3+4 findings in patient 3 in the left midgland (contralateral to the ¹³C target) was small-volume (1 out of 4 cores in the sextant, <5% involvement per core) and only detected by systematic biopsy. The k_PL value per lesion was calculated from the maximum voxel. CAPRA, Cancer of the Prostate Risk Assessment; PSA, prostate-specific antigen.

TABLE 1.

Summary of the clinical characteristics and biopsy findings from 3 representative patients in this study. (The remaining 2 patients can be found in Supporting Information Table S1). All integrated ¹³C-¹H mpMRI studies and the associated biopsies were safe and successful without adverse events. The median number of ¹³C targets was 1 (range: 1–2) per patient, measuring 1 cm (range:0.6–1.9) in diameter, with a median k_PL of 0.0319 s⁻¹ (range: 0.0198–0.0410). Low-grade cancer involvement was found in the cores corresponding to the ¹³C targets in 3 out of 5 patients. These findings were consistent with clinically low- to intermediate-risk disease

Patient No.	Patient 1	Patient 2	Patient 3
PSA	9.6	8.4	4.2
UCSF CAPRA score	1	1	3
Risk group	Low	Low	Intermediate
No. of 13C targets	1	1	1
kPL at 13C target(s−1)	0.0378 (σ = 0.0050)	0.0198 (σ = 0.0018)	0.0325 (σ = NA)
No. of (+) cores/total cores in targeted + systematic biopsy	4/17	5/18	10/18
Overall grade	3 + 3	3 + 3	3 + 4
No. of (+) 13C cores/total 13C cores	1/2	1/2	0/3
13C core grade	3 + 3	3 + 3	Benign

¹H, proton; ¹³C, carbon-13; k_PL, pyruvate-to-lactate conversion rate; mp MRI, multiparametric MRI; NA, nonapplicable; PSA, prostate-specific antigen; UCSF CAPRA score, University of California, San Francisco–Cancer of the Prostate Risk Assessment score.

3.3 |. Correlation between k_PL and histopathologic findings from biopsy

Overall, 1 patient (patient 3) had Gleason 3+4, and 4 patients (patient 1,2,4,5) had 3+3 disease (number of biopsy cores positive for cancer per patient, median: 4/17, range: 3/19–10/18), combining pathological findings from standard and ¹³C research-targeted biopsies (Table 1) (Figure 5) (Supporting Information Table S1). On a per patient basis, the maximum involvement of any core was 16%–52% (median: 16%). The cores sampled from ¹³C research targets were Gleason 3+3 in 4 patients (patient 1,2,4,5), with median 16% involvement (range: 1%–16%). One patient (patient 5) had atypical small acinar proliferation and high-grade prostatic intraepithelial neoplasia among the ¹³C cores in 1 target, and the ¹³C target in another patient (patient 3) contained benign prostate tissue.

FIGURE 5.

HP ¹³C-k_PL targets from the 5 cases summarized in Table 1 and Supporting Information Table S1

Figure 6 illustrates a representative case (patient 1) who underwent fusion plus systematic biopsies with a ¹³C research target (k_PL = 0.0378 s⁻¹) at the left mid-apex peripheral zone and a ¹H mpMRI target (PIRADS 4) at the right mid-base transition zone. Pathological diagnosis of the tissue sample from the ¹³C target was Gleason 3+3 cancer (16% involvement, 1/2 cores), whereas that from the ¹H-MRI target was described in the histology report as “rare atypical glands.” Systematic biopsy found 3/12 cores with low volume 3+3 disease. Altogether, patient 1 had CAPRA score of 1, consistent with the clinically low-risk diagnosis.

FIGURE 6.

(A) Shows an example of an integrated HP-¹³C research and standard ¹H mpMRI study (patient 1) including key multiparametric T₂-weighted, diffusion, and k_PL images identifying the biopsy target. One ¹H target (PIRADS 4) was identified at right mid-base transition zone and one ¹³C research target (k_PL = 0.0378 s⁻¹) at left mid-apex peripheral zone, as indicated by the arrows. (B) ¹³C and ¹H mpMRI biopsy targets as drawn by an experienced abdominal radiologist. Pathological diagnosis of the tissue sample from the ¹³C target was Gleason 3+3 cancer (16% involvement, 1/2 cores), whereas that from the ¹H-MRI target was described in the histology report as “rare atypical glands”

Taken together with other clinical biomarkers, the biopsy findings in these 5 patients were consistent with clinically low- to intermediate-risk disease (summarized in Table 1), indicating that these patients are appropriate candidates for active surveillance. Four patients continued active surveillance after the study, whereas 1 patient later elected to undergo definitive treatment.

4 |. DISCUSSION

We investigated for the first time the safety and technical feasibility of integrating a rapid, 1-min hyperpolarized ¹³C-pyruvate MRI acquisition into standard-of-care ¹H mpMRI exams for guiding TRUS fusion prostate biopsies. No HP ¹³C MRI- or biopsy-related adverse events were reported in this initial cohort, which is in agreement with the excellent safety record of 600+ hyperpolarized ¹³C MRI studies conducted worldwide thus far on patients and volunteers, as well as that of the active surveillance biopsy procedure at our institution and many other centers globally.^29,30 These results support future investigations focused on determining whether HP ¹³C-¹H mpMRI could benefit men who are either candidates for or undergoing active surveillance of prostate cancer.

Using the existing infrastructure, we incorporated the rapid HP ¹³C-pyruvate scan and the associated metabolic biomarker k_PL into the routine prostate MRI workflow, for which the major components includes image postprocessing, radiology read with lesion targeting, MR-guided TRUS fusion biopsies, and finally pathology evaluation. The integrable nature of this approach proved effective to improve efficiency and minimize additional effort.

The new ¹³C targeting feature took advantage of the existing overlay and lesion outlining functions on a commercial, out-of-the-box prostate mpMRI processing, and targeting platform (Dynacad, Philips Invivo Corp.). This not only enabled easy deployment of the ¹³C targeting capabilities on any PACS workstation within our radiology network, without the burden of additional software installations or modifications, but also allowed directly exporting the ¹³C targets from the PACS/Dynacad (Philips Invivo Corp., Gainsville FL) to the fusion biopsy platform (UroNav, Philips Invivo Corp.) in the urological oncologists’ offices, along with the ¹H mpMRI targets. We envision the same rationale and workflow are generalizable to other commercial targeting and fusion biopsy platforms^31,32 in a vendor-independent fashion. Our approach is presumably suitable for either transrectal or transperineal biopsy techniques because most commercial platforms support both by default.³³

The ¹³C research biopsy results were readily incorporated in the pathology report, and the pathological information of the ¹³C biopsy cores, including the Gleason scores, total percentage of tumor involvement, percentage of Gleason ≥4 pattern, and presence of adverse pathological features, were directly factored into the patients’ clinical risk calculations, such as the UCSF-CAPRA score utilized at our institution. This paves the way for easy incorporation into routine clinical practice in the future.

Five patients were enrolled to test this new approach in a pilot, technical feasibility study. Interestingly, 4 out of 6 ¹³C-k_PL targets did not correlate to a PIRADS lesion on ¹H mpMRI on a per-lesion basis. The discrepancy is consistent with the knowledge that HP ¹³C MRI offers unique, complementary information based on prostate tumor metabolism in addition to the anatomical and functional features provided by ¹H mpMRI. It also highlights the need to investigate, in a larger cohort, whether a HP ¹³C-¹H mpMRI protocol may overcome the known limitation in sensitivity with conventional prostate MRI for detection of occult but clinically significant tumors in the surveillance setting.³⁴

Our study design substituted systematic biopsy cores with ¹³C-targeted biopsy cores in the same sextant. The rationale was to reduce oversampling bias as a confounder when comparing the diagnostic accuracy between men who received fusion biopsies, including the ¹³C targets, versus those who only received standard-of-care targeted and systematic biopsies. Our approach was consistent with that taken by the Active Surveillance Magnetic Resonance Imaging (ASIST) trial,³⁵ also designed to evaluate the utility of mpMRI-guided biopsy. An additional benefit was to fulfill the technique development aim for ¹³C-guided fusion biopsy without introducing additional biopsy-related morbidity to the men who participated in this technical feasibility study.

Distinct from standard mpMRI series, the ¹³C-k_PL series was labeled nondiagnostic on PACS. On Dynacad (Philips Invivo Corp.), the k_PL target lesions were labeled as C13 lesion #, setting them apart from the ¹H mpMRI targets, which were named lesion 1, lesion 2, etc. Using separate labels reduced equivocal nomenclature and improved communications between imaging researchers and the patients’ multidisciplinary care team by making the distinction between ¹³C research imaging results and associated target lesions from those of ¹H mpMRI.

Whereas this study successfully demonstrated the safety and technical feasibility of guiding fusion prostate biopsies using the HP ¹³C MRI-derived k_PL biomarker, a few limitations and challenges need to be acknowledged. First, a consensus in the field of prostate mpMRI is trending away from the use of endorectal coils. Similar to current conventional ¹H MRI coils, new flexible ¹³C array coils are becoming commercially available to provide wider spatial coverage and higher SNR. Combining with the recent developments in ¹³C parallel imaging³⁶ and denoising algorithms,^37,38 these advancements will likely obviate the need of endorectal coils for future ¹³C-pyruvate prostate studies.

Second, the k_PL cutoff value (0.02 s⁻¹), representing the dichotomy between high-grade prostate adenocarcinoma and low-grade tumor/benign tissue, was derived from the histopathology of a high-risk cohort who underwent radical prostatectomy.¹⁷ This high-risk cohort likely possesses quite distinct underlying biology from the lower-risk, active surveillance population who may benefit the most from HP ¹³C MRI. Whether the same k_PL cutoff value is appropriate for detecting occult high-grade disease in the surveillance population thus needs to be further explored and validated. Additionally, developing and testing a PIRADS-like grading schema for HP ¹³C would benefit future clinical research but will require a broader evaluation in larger patient populations, as well as input from research radiologists on the methods developed in this project.

Although replacing the systematic cores with ¹³C-k_PL cores reduces oversampling bias, a potential downside would be the missed opportunity to determine whether the standard 12–14 core systematic TRUS biopsy would have also found the 3+3 disease as the ¹³C-guided biopsy in the same sextant on an individual-patient basis.

The primary factor affecting biopsy accuracy is the imperfect registration of MR-TRUS software fusion and mechanical deflection of the biopsy needles, which can similarly affect ¹H mpMRI targeted biopsies.³⁹ Therefore, biopsies are not considered a gold standard like postprostatectomy step-section histopathology is.

Finally, this technical feasibility study was neither designed nor powered to calculate the sensitivity/specificity of HP ¹³C MRI-guided biopsy, and it would not be appropriate to report the clinical utility or biological significance results given the limited data and lack of a gold standard. The clinical question of whether ¹³C MRI improves diagnostic accuracy of prostate cancer needs to be tested in a future larger-scale clinical trial. We believe such a trial is feasible given multiple NCI-designated cancer centers are now equipped with HP ¹³C MRI capabilities and are either already performing or planning to conduct ¹³C prostate cancer research.^40,41 Assuming HP ¹³C MRI is proven to increase detection of clinically significant prostate cancer, whether this improvement ultimately translates into better disease-specific outcome will require long-term follow-up studies.

5 |. CONCLUSION

This technical development study demonstrated the feasibility of adding HP ¹³C-pyruvate MRI to guide TRUS fusion prostate biopsies. HP ¹³C MRI biomarkers were integrated into the diagnostic mpMRI workflow, complete with identification of ¹³C research targets and sampling of these targets in fusion biopsies. These initial results support future studies in a larger cohort of patients to evaluate the role of HP ¹³C MRI–guided targeted biopsy for improving prostate cancer risk stratification.

Supplementary Material

mrm29399-sup-0001-Supino.pdf

FIGURE S1. (A) An example case showing the comparison of k_PL in the ¹³C targeted lesion versus segmented prostate outside of the lesion. Pathological diagnosis of the biopsy tissue was Gleason 3 + 3 tumor with 16% involvement. (B) k_PL dichotomy between pathologist-defined low-grade prostate cancer (PCa) (Gleason ≤3 + 4), and high-grade prostate cancer (Gleason score≥4 + 3). *p = 0.034; **p = 0.0003. The recommended k_PL threshold = 0.02(s⁻¹) used in our study was corrected for the different MR sequence echo times between the EPSI acquisition in the cited reference versus the EPI in our study.²⁴ Figure reproduced with permission¹⁷

TABLE S1. Summary of the clinical characteristics and biopsy findings from Patient 4 and 5

mrm29399-sup-0002-VideoS1.mp4

VIDEO S1. This video illustrates the protocol to create ¹³C-k_PL/T₂ fusion overlays. The new ¹³C biopsy targeting feature takes advantage of the existing overlay and lesion outlining functions on a commercial, out-of-the-box prostate mpMRI processing and targeting platform (Dynacad). This not only enabled easy deployment of the ¹³C biopsy targeting capabilities on any PACS workstation within our radiology network - without the burden of additional software installations or modifications, but also allows directly exporting the ¹³C targets from the PACS/Dynacad to the fusion biopsy platform (UroNav)