A Latent Cardiomyocyte Regeneration Potential in Human Heart Disease

Heart tissue was procured from the European Homograft Bank (Brussels, Belgium); Medical University of Graz (Austria); KI Donatum, Karolinska Institute Stockholm, (Sweden); University of Lund (Sweden); institutions comprising the Utah Transplantation Affiliated Hospitals Cardiac Transplant Program (ie, University of Utah Health Science Center, Intermountain Medical Center and the Veterans Administration Salt Lake City Health Care System), Spectrum Health Universal Biorepository (Utah) and Gift of Life (Michigan). Ethical permission for these studies was granted by the regional ethics review board in Stockholm (DNR 2005/1029-31/2 and DNR 2010/313-31) and the Ethikkommission an der TU Dresden (BO-EK-50022020), and informed consent was provided by all individuals. Nondiseased human heart tissues were obtained by KI Donatum after informed consent from the individual or next of kin from the Swedish National Department of Forensic Medicine. Relevant clinical parameters are listed in Tables S1 and S2. Tissue from the left ventricle harvested at the time of heart transplantation, dissected, and stored at −80 °C until further processing. Tissue samples were processed in 3 ways: (1) nuclear isolation and fluorescence-activated cell sorting (FACS) sorting for ¹⁴C analysis, (2) paraffin embedding for subsequent histological analysis, or (3) isolation of intact cardiomyocytes from fixed tissue samples.

Large frozen heart pieces of the free ventricular wall (minimum, 10 g) were dissected and dissociated in a blender in lysis buffer (0.32 M sucrose, 10 mM Tris-HCl pH=8.0, 5 mM CaCl₂, 5 mM MgAc, 2 mM EDTA) for 10 minutes. The tissue was further homogenized with a T-25 Ultra-Turrax probe homogenizer (IKA Germany) at 24 000 rpm for 10 s and afterward dounced with a type A pestle in a 40-mL glass douncer, applying 8 strokes. The nuclear isolate was filtered through 100- and 60-µm nylon mesh cell strainers and sedimented at 500× g for 10 minutes. The pellet was dissolved in sucrose buffer (2.0 M sucrose, 10 mM Tris-HCl, pH 8, 5 mM MgAc), and layered onto 10 mL of sucrose buffer in a BSA-coated centrifuge tube. The samples were spun down at 26 500× g for 60 minutes in a Beckman Avanti Centrifuge (Beckman Coulter). The nuclei pellets were dissolved in nucleus storage buffer (0.44 M sucrose, 10 mM Tris-HCl pH 8.0, 70 mM KCl, 10 mM MgCl₂, 2 mM EDTA). All steps were performed at 4 °C.

A polyclonal rabbit anti-PCM1 ([pericentriolar material 1] Sigma Aldrich, HPA023374) antibody, conjugated to Alexa Fluor 488 or without conjugation to a fluorochrome, was added to the sample in a 1:800 dilution and incubated overnight. The next day, samples of isolated nuclei were filtered through 40 µm cell strainers (BD Bioscience) and labelled with DNA stain DRAQ5 (ThermoFisher, 62251, 1:1000) for ploidy analysis. Samples that were not processed with a primary antibody conjugated to a fluorochrome were incubated with a secondary anti-rabbit Alexa Flour 488 (1:1000) for 1 hour. Negative controls were incubated without primary antibody. Subsequently, the sample was loaded to the flow cytometer (FACS Aria II), single nuclei were identified by forward scatter height/forward scatter area gating, and cardiomyocyte nuclei were defined by gating the PCM1⁺ nuclei. After nuclei sorting, the populations were reanalyzed for purity. The sorted nuclei were spun down at 500× g for 10 minutes, dissolved in PBS, and kept at −20 °C until further processing. All steps were performed at 4 °C. The purity of the PCM1⁺ sorted fraction (n=52) was (median) 95.1% (interquartile range [IQR], 92.2% to 96.7%; Figure S2E). If the sorting purity was <100%, the DNA purity of the sample was calculated according to the following equation:

Where x is the sample being corrected, f_x is the nuclei purity measured in the FACS, f_1…n are the fractions of the n contaminating populations, c_1…n are the DNA contents of the contaminating nuclei (eg, 4n: c₁=4), reflecting that polyploidy nuclei contain more DNA.

DNA extraction from flow cytometry sorted PCM1⁺ nuclei of more than one million nuclei was performed under clean-room conditions; 500 µL of DNA lysis buffer (1% SDS, 5 mM EDTA-Na₂, 10 mM Tris-HCl pH 8) and 6 µL of proteinase K (20 mg/mL; Life Technologies, EO0491) were added to each sample of sorted nuclei. The samples were incubated at 64 °C overnight. After adding half the volume of 5 M NaCl, the samples were vortexed for 15 seconds and spun down at 16 000× g for 3 minutes. The supernatants were transferred to 10-mL glass tubes, 3× volumes of 96% ethanol were added, and the glass tubes were inverted several times until a DNA precipitate became visible. The DNA pellet was washed in DNA washing buffer (70% ethanol, 0.5 M NaCl) 3 times for 15 minutes each. The washed DNA pellets were dissolved in 200 µL of sterile double distilled water (Gibco) and allowed to dry at 64 °C overnight. The dried DNA pellet was dissolved in 500 µL of sterile water overnight. The sample was analyzed for DNA concentration and purity by measuring the absorbance at 260 nm and 280 nm with a spectrophotometer (NanoDrop 1000, Thermo Scientific). Sample DNA abundance was median 42.9 µg DNA (IQR, 30.0–62.1 µg; n=52; Figure S2F).

Accelerator mass spectrometry measurements were performed blind to the identity of the samples and as described previously.¹⁶ Purified DNA samples (Figure S2F) were lyophilized to dryness. Excess copper (II) oxide was added to each dry sample, and the tubes were evacuated to 2.10^-3 mbar and sealed with a high-temperature torch. The tubes were placed in a furnace set at 900 °C for 3.5 hours to combust all carbon to carbon dioxide. The carbon dioxide was then cryogenically purified, trapped, and reduced to graphite in the presence of an iron catalyst in individual reactors at 550 °C for 6 hours. Graphite targets were measured at the Department of Physics and Astronomy (Ion Physics, Uppsala University).¹⁷ Stable Isotope ratio, δ¹³C, was measured for each sample to correct for any isotopic fractionation. Strict laboratory practice was implemented to minimize stray background carbon contaminating the samples. Corrections for background carbon contamination introduced during sample preparation were made as described.^{17 14}C data are reported as decay-corrected Δ¹⁴C. The measurement accuracy was in the range of 1% to 3% (2σ).

Wheat germ agglutinin (WGA) staining was performed on 5-µm paraffin sections prepared from the formaldehyde fixed tissue samples. Sections were deparaffinized using xylene and ethanol and incubated with WGA, Alexa Fluor 488 Conjugate (1:1000; ThermoFisher, W11261, and diluted in PBS. Sections were washed 3 times with PBS and mounted with ProLong Gold Antifade Reagent containing DAPI (Life Technologies, P36935). Cardiomyocyte cross-sectional area was determined using CellProfiler 2.0¹⁸; analysis pipeline is provided as a data file (https://zenodo.org/record/7560219).

To analyze mtDNA content total DNA was extracted from the tissue samples for processing with Takara Human mtDNA Monitoring Primer Set (category No.7246). Tissue biopsies were taken before the start of the nuclei isolation procedure, using a 6-mm wide disposable biopsy punch (Kai Medical, No. 48601) and homogenized by the Ultra-Turrax homogenizer at 24 000 rpm for 10 seconds in 1 mL of lysis buffer (0.32 M sucrose, 5 mM CaCl2, 5 mM magnesium acetate, 2 mM EDTA, and 10 mMTris-HCl, pH 8.0). DNA was isolated using 200 µL of the lysate, with the QIAamp DNA Blood Mini Kit (Qiagen, 51104) according to manufacturer instructions. DNA concentration and purity was determined using NanoDrop. Quantitative real-time PCR reactions were set up in a total volume of 25 μL using the Takara Human mtDNA Monitoring Primer Set (Takara, 7246) to quantify the relative number of copies of human mtDNA with respect to the nuclear genomic DNA (nDNA). Reactions were set up according to manufacturer protocol using 12.5 μL of SYBR Green master mix (Thermo Fisher, No. 4309155) and 10 ng of template DNA. Thermocycling was set up in the StepOne Real-Time PCR System (Applied Biosystems, No. 4376357).

Tissue samples from the left ventricular wall and apical tissue obtained at the time of LVAD implantation were cut into 0.5- to 1-mm cubes and fixed in 3.7% formaldehyde solution for 100 minutes at room temperature. Subsequently, tissue blocks were washed 3 times with PBS for 30 minutes and then digested in a buffer consisting of collagenase B (Roche, 11088815001; 3.6 mg/mL) and collagenase D (Roche, 11088858001; 4.8 mg/mL) in PBS for 24 hours at 37 °C with gentle rotation. Isolated cardiomyocytes were gently centrifuged at 1× g, and the supernatant was removed before resuspending in PBS. For several samples, this procedure was repeated for an additional 24 hours on the remaining tissue.

Isolated cardiomyocytes were allowed to settle for 15 minutes and were resuspended in 500 µL of blocking buffer with primary antibodies Mouse-α-Actinin,1:500 (Sigma Aldrich, A7811) and Rabbit-Connexin-43 (1:1000 [Sigma Aldrich, C6219]) for 30 minutes. Cardiomyocytes were washed once with PBS for 15 minutes and incubated with secondary antibody solution Alexa Fluor 555 anti-mouse, 1:1000 (Jackson Immuno Research, 711-546-152) and Alexa Fluor 488 anti-rabbit, 1:1000 (Abcam, ab150110) and Hoechst 33342 (1:1000 [ThermoFisher, 62249]) for 30 minutes. Cardiomyocytes were washed with PBS and resuspended in a small volume of PBS and pipetted onto a glass slide. Cardiomyocytes on the glass slide were allowed to settle and adhere before mounting (ProLong Gold antifade reagent; ThermoFisher, P36930). All imaging was performed on a Keyence BZ-X800E using 20× objectives with appropriate dichroic filters. Exposure times were chosen that allowed capturing of the full dynamic range without overexposure and were similar for all acquired images. At least 30 fields of view were recorded per sample in a nonoverlapping manner.

To identify nuclei in the intact cardiomyocytes isolated from fixed tissue and measure their integrated intensities, a CellProfiler 4.0¹⁹ pipeline was set up (https://zenodo.org/record/7560219). Nuclei identification was performed using the identify primary objects module with an adaptive thresholding method to account for background variances. The α-Actinin image was used to exclude noncardiomyocyte nuclei that do not fall within the α-Actinin positive areas. Further filtering by several neighbors and solidity was chosen to exclude large clumps of nuclear stains and irregular shapes, which indicated bad segmentation or accumulation of staining that did not resemble individual nuclei. Area and shape descriptors, as well as intensity descriptors, were exported for further analysis and integrated fluorescence intensity of the Hoechst stain was plotted as histograms to determine ploidy thresholds manually (Figure S2G through S2I). These thresholds were then used to overlay colored outlines on the overlay pictures for further analysis. Evaluation of the ploidy levels in mononucleated and binucleated cells was performed manually based on the overlay pictures, considering cardiomyocyte shape and border morphology using connexin-43 staining. Per sample, an average of 168 cardiomyocytes were analyzed with a minimum of 50.

Twenty-eight of 52 patients with advanced heart failure requiring circulatory support with continuous flow LVAD were prospectively enrolled at one of the institutions comprising the Utah Transplantation Affiliated Hospitals Cardiac Transplant Program (ie, University of Utah Health Science Center, Intermountain Medical Center, and Veterans Administration Salt Lake City Health Care System). Patients requiring LVAD support because of acute heart failure (eg, acute myocardial infarction, acute myocarditis, or postcardiotomy cardiogenic shock) were excluded. Serial echocardiographic assessments were performed prospectively within 1 month before LVAD implantation and at 1, 2, 3, 6, 9, and 12 months after LVAD implantation based on a previously described protocol.²⁰ Patients were categorized as responders and nonresponders based on the absolute left ventricular ejection fraction (LVEF) change from pre–LVAD to post–LVAD implantation. Patients with an absolute LVEF increase ≥5% within 1 year on LVAD support were defined as responders.

We made use of a previously published single-cell sequencing data set of end-stage heart failure by Amrute et al.²¹ We downloaded the list of differentially expressed genes from Amrute et al (Table S7 by Amrute et al) for the comparison of “RRpost vs preLVAD HF.”²¹ These lists contain all differentially expressed genes (adjusted P value<0.05). Amrute et al used the DESeq2 package for this analysis, and the P values attained by the Wald test are corrected for multiple testing using the Benjamini and Hochberg method by default. In these lists, we then marked all genes that are members of the MSigDB v2023.2 Hallmark DNA repair gene set as DNA repair associated genes. Only 4 of the 150 Hallmark DNA repair genes were differentially expressed.

Values are given as mean±SD for normal distributed data and median±IQR for non-normal distributed data unless stated otherwise. Statistical tests were performed, and graphs were generated using Graphpad Prism Version 10 and SigmaPlot 13. One-way ANOVA was performed when the data had a normal distribution (Shapiro–Wilk) and the variances were equal (Brown–Forsythe). Kruskal–Wallis 1-way ANOVA on ranks was performed when any of the above conditions were not met. Appropriate post hoc multiple comparison tests were performed as indicated in the figure legends. P<0.05 was considered statistically significant. The Mann-Whitney rank sum test was used to compare 2 groups that were not normally distributed. Linear regression analysis was performed to calculate R and whether the slope is significant from 0. Compositional data underwent an isometric log-ratio transformation to address its compositional nature and facilitate subsequent statistical analysis.²² After the isometric log-ratio transformation, a principal component analysis was conducted to reduce the dimensionality of the data and identify patterns of variation among the groups. Only the first principal component was used for further analysis, reflecting the most significant axis of variation. The ANOVA was then employed to detect significant differences in first principal component scores across the 3 groups. To elucidate the specific differences identified by the ANOVA, a Tukey honestly significant difference post hoc test was performed, facilitating pairwise comparisons between the groups. The analysis was conducted using R 4.3.3 (R Core Team, 2024), leveraging the following packages: compositions for the isometric log-ratio transformation, multcomp for conducting Tukey honestly significant difference post hoc test, readxl for importing the data set, and tidyverse for data manipulation and visualization.

National Institutes of Health principles and guidelines for reporting preclinical research were followed. Experimenters were blind to group assignment and outcome assessment. For some analyses, a random subset of our patient population was selected (Table S3). No randomization was performed, as the groups were disease categories from human samples. Patient demographical details can be found in Tables S1 and S2. No sample size calculation was performed. Sample size was governed by tissue availability.

All experimental data reported in this article will be shared by the lead contact upon request. All original code has been deposited at Zenodo and is publicly available (https://zenodo.org/record/7560219). Any additional information required to reanalyze the data reported in this article is available from the lead corresponding author upon request.

Cardiomyocytes often exit the cell cycle prematurely and become polyploid without forming daughter cells. We determined nuclear ploidy in cardiomyocytes of patients with failing hearts diagnosed with NICM and ICM (Table S1). Nuclei were isolated from tissue samples and cardiomyocyte nuclei identified using flow cytometry by their positive staining for PCM1 (mean±SD, 21.3±14.5% of all nuclei [Figure 1A; Figure S2]).^2,23,24 Application of a DNA stain (DRAQ5) enabled the determination of their nuclear DNA content (Figure 1B). We found a shift toward higher nuclear ploidy levels in both NICM and ICM (Figure 1C) compared with healthy subjects (values taken from Bergmann et al²). As a measure of nuclear ploidy, we defined the degree of nuclear ploidy as a percentage in relation to 100% of a purely diploid (2n) reference sample.^2,3

We observed a total increase in cardiomyocyte nuclear ploidy level from 176.9% (median [IQR, 169.8–196.9%]) in healthy hearts to 291.7% (250.4–341.4%) in NICM and 221.1% (202.1–255.0%) in ICM (Figure 1D). The increase in nuclear ploidy was not linked to the age of the patient (NICM, R=0.26, P=0.26; ICM, R=0.49, P=0.11; Figure 1E).

In a second approach, we isolated intact single cardiomyocytes and performed image cytometry to determine the number of nuclei per cardiomyocyte and their ploidy status (Figure 1F; Figure S2G and S2I). Using a sample set of healthy adults, we found 19.3±3.0% of cardiomyocytes to be binucleated (Figure 1G; Table S4), in agreement with previous studies, which also showed that this ratio remains unchanged in a healthy heart throughout the human lifespan.^2,25,26 In contrast, we observed that cardiomyocytes isolated from NICM and ICM hearts showed significantly higher levels of binucleation compared with healthy hearts: 34.6±5.1% in NICM and 35.2±5.5% in ICM patients. Similarly, there was no correlation between the age of the patient and the ratio of binucleated cardiomyocytes (NICM, R=0.13, P=0.52; ICM, R=0.04, P=0.92; Figure S2D).

Furthermore, we determined individual ploidy levels within mononucleated and binucleated cardiomyocytes. This analysis revealed that the proportion of mononucleated diploid cardiomyocytes (1×2n) is smaller in NICM (median, 10.35% [IQR, 4.6–14.5%] and ICM (7.8% [6.3–25.3%]) compared with healthy hearts (24.0% [17.0–39.0%]; Figure 1H and 1I). Although both nuclear ploidy and multinucleation increase in NICM and ICM, the distribution of ploidy levels within the mononucleated and binucleated cardiomyocytes is independent of the number of nuclei per myocyte (Figure S2K). These measurements by image cytometry are consistent with the extent of nuclear ploidy determined by flow cytometry, shown by Bland–Altman comparison and linear regression analysis (Figure S2H through S2J). In conclusion, both demonstrate dramatically increased cardiomyocyte DNA content in diseased failing human hearts from both ICM and NICM patients.

We isolated cardiomyocyte nuclei (PCM1⁺) from NICM (n=16) and ICM patients (n=8), between 18 and 68 years of age (median, 51.5 years; Figure S2A) by FACS (Figure S2B and S2C) and determined the genomic ¹⁴C concentrations by accelerator mass spectrometry (Supplemental Methods, Tables S1 and S2; Figure 2A). The ¹⁴C amounts measured in NICM and ICM samples deviate more from the atmospheric ¹⁴C levels at the time of birth, than the values from healthy subjects,² suggesting that there was more cardiomyocyte DNA synthesis in these patients. To better classify this divergence, we determined the average ¹⁴C age of the cardiomyocytes’ genomic DNA (Figure 2B; Figure S1A and S1B), which provides an initial estimation of cellular ages under the assumption that all cells were generated at the same time.²⁷ In healthy subjects, the genomic ¹⁴C age of cardiomyocyte DNA was an average of 6 years (median [IQR, 4.7–6.7 years]) younger than the subject itself (Figure 2C; data set taken from reference 2). In contrast, we found the genomic ¹⁴C age of NICM cardiomyocytes to be 12.4 years (median [IQR, 8.5–16.1 years]), and of ICM cardiomyocytes to be 11.1 years (IQR, 9.2–13.6 years) younger than the individual, respectively. These observations demonstrate increased levels of DNA synthesis in cardiomyocytes from NICM and ICM patients compared with healthy controls.

We defined a population balance equation for ¹⁴C concentration structured cell populations, which we used to predict ¹⁴C concentration dynamics from cell renewal rates (see Mathematical Methods) using a Bayesian inference framework.²⁷ The cell renewal rates are allowed to change in time, which enables us to model different rates before disease onset, during the disease, and after treatment with LVAD. For NICM and ICM patients, we assumed that polyploidization and multinucleation proceed physiologically until disease onset and increased thereafter until tissue procurement. For this assumption, we applied the pathological ploidy levels individually measured for all ¹⁴C dated samples (Figure 1D) and levels of binucleation determined for each population as depicted in Figure 1G.

To establish the rate of cardiomyocyte renewal occurring specifically during the disease phase, we defined 2 phases of cardiomyocyte renewal: before and after the disease onset (first occurrence of heart failure symptoms). In this 2-phase model, we assumed that the renewal rates before disease onset are the same as in healthy subjects (0.55% per year). The fitted renewal rates after disease onset resulted in much lower median renewal rate values of 0.03% per year for NICM patients and 0.01% per year for ICM patients, corresponding to an 18- and 50-fold reduction, respectively, compared with the healthy heart (Figure 3A).

To test for the possibility that changes in cardiomyocyte renewal could be part of disease ontogeny in NICM patients, we used our simplified model assuming a constant lifelong cardiomyocyte renewal (Supplemental Methods). We find that the distribution of cardiomyocyte renewal is wider with a similar median in NICM compared with healthy hearts (0.55% per year in healthy hearts and 0.59% per year in NICM), suggesting that there is no overall early loss of the capacity to renew in the NICM group (Supplemental Methods).

The fact that cardiomyocyte renewal is greatly reduced in heart failure suggests that the increased cardiomyocyte DNA synthesis, as determined by changes in genomic ¹⁴C levels in failing hearts, is almost exclusively related to polyploidization and multinucleation. (Figure 2B and 2C). To quantify this, we inferred total DNA synthesis from genomic ¹⁴C concentration in cardiomyocytes and examined the contribution of polyploidization, multinucleation, and cardiomyocyte renewal attributable to proliferation to total DNA synthesis. Applying our 2-phase mathematical model, we found that most of the DNA synthesis at the timepoint 1 year before disease onset was associated with proliferation-based cardiomyocyte renewal (median, 90.0% [IQR, 70.5–93.0%]; Figure 3C), whereas this rate dropped dramatically to values of 0.2% (median [IQR, 0.1–0.4%] and 0.2% (median [IQR, 0.1–0.7%]) at the timepoint of collection during the disease phase of NICM and ICM hearts, respectively. In those samples, most of the DNA synthesis was indeed associated with polyploidization (NICM median, 92.5% [IQR, 89.8–94.3%]; ICM 85.5% [80.5–93.3%]) and multinucleation (NICM, 7.5% [5.0–7.5%]; ICM, 14.0% [6.0–19.5%]; Figure 3C).

In summary, polyploidization and multinucleation explain >99.5% of the ¹⁴C incorporated into the DNA of failing cardiomyocytes, whereas cardiomyocyte renewal drops drastically to minimal levels.

Having observed the dramatically reduced cardiomyocyte renewal rate of the failing heart, we inquired whether this could be rescued by mechanical unloading and circulatory support. We analyzed advanced heart failure patients (including NICM and ICM) receiving LVAD therapy for 3 to 43 months (Figure S4A and S4B). Failing hearts not supported with an LVAD were used as the reference group (non-LVAD). It has been shown that in advanced heart failure patients, LVAD mechanical unloading and circulatory support can improve cardiac function to varying degrees.^8,20 Based on these LVAD studies as well as investigations of other heart failure therapies,^28,29 we categorized all patients with an absolute increase of LVEF >5% as responders. To compare cardiomyocyte hypertrophy, we measured their cross-sectional area in these subgroups (Figure S4C). Responders showed smaller cardiomyocyte cross-sectional areas (832.8±307 µm²) than patients with no improvement in LVEF ([nonresponders] 1123±322 µm²; Figure S4D). Moreover, we studied mitochondrial abundance quantified by the relative copy number of human mitochondrial DNA by real-time PCR and normalized to genomic DNA copy numbers. Normalized mitochondrial DNA was lower in LVAD responders (median, 1984 [IQR, 1143–3736]) than in the LVAD nonresponder group (4025 [2918–5233]; P=0.02]) and non-LVAD patients (3487 [2821–5962]; P=0.01]; Figure S4E). Reversal of cardiomyocyte hypertrophy and decrease in mitochondrial load are 2 parameters of reverse remodeling that we have previously reported to be associated with mechanical unloading and circulatory support.³⁰

In contrast to previously published work,^13,31 we did not detect changes in nuclear ploidy or multinucleation in mechanically unloaded hearts using flow cytometry and image cytometry (Figure S4F and S4G). Neither responders nor nonresponders showed significant differences in these parameters compared with non-LVAD advanced heart failure patients.

Next, we determined the ¹⁴C concentration in cardiomyocyte DNA in LVAD responders (n=15) and nonresponders ([n=13] see Methods; Tables S1 and S2) to establish their ages and renewal rates compared with non-LVAD patients ([n=24] Figure 4A). ¹⁴C concentrations of responders deviate more from atmospheric levels than non-LVAD patients. To classify potential fundamental differences in cardiomyocyte age within the three subgroups, we determined the genomic ¹⁴C ages as a first estimate and plotted these against the age of the patient (Figure 4B). The group of responders was shown to differ the most from the no turnover prediction line. We compared these average genomic ¹⁴C DNA ages with the ages of the subjects and observed that it was 12.6±4.6 years fewer in the non-LVAD patients and 13.9±3.6 years for nonresponders, compared with 19.3±6.0 years in responders (Figure 4C).

To further assess the impact of LVAD treatment on cardiomyocyte renewal during the disease process, we applied the aforementioned 2-phase model of cardiomyocyte renewal. The fitted renewal rates after disease onset resulted in median renewal values of 0.03% per year for non-LVAD and 0.02% for LVAD nonresponder patients; in contrast, renewal values in LVAD responders were 3.1% per year (Figure 5A). In responders, 29.7% (median [IQR, 8.5–54.5%]) of all cardiomyocyte DNA synthesis can be attributed to cell renewal, contrasting with only 0.26% (0.1–0.6%) and 0.23% (0.01–1.0%) in nonresponders and non-LVAD patients, respectively; almost all DNA synthesis is attributed to polyploidization and binucleation (Figure 5B). This suggests a fundamental difference in the regenerative capacity in patients with advanced heart failure exhibiting cardiac functional improvement during LVAD mechanical unloading.