Utility of new image-derived biomarkers for autosomal dominant polycystic kidney disease prognosis using automated instance cyst segmentation

Abstract

New image-derived biomarkers for patients affected by autosomal dominant polycystic kidney disease are needed to improve current clinical management. The measurement of total kidney volume (TKV) provides critical information for clinicians to drive care decisions. However, patients with similar TKV may present with very different phenotypes, often requiring subjective decisions based on other factors (e.g., appearance of healthy kidney parenchyma, a few cysts contributing significantly to overall TKV, etc.). In this study, we describe a new technique to individually segment cysts and quantify biometric parameters including cyst volume, cyst number, parenchyma volume, and cyst parenchyma surface area. Using data from the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) study the utility of these new parameters was explored, both quantitatively as well as visually. Total cyst number and cyst parenchyma surface area showed superior prediction of the slope of estimated glomerular filtration rate decline, kidney failure and chronic kidney disease stages 3A, 3B, and 4, compared to TKV. In addition, presentations such as a few large cysts contributing significantly to overall kidney volume were shown to be much better stratified in terms of outcome predictions. Thus, these new image biomarkers, which can be obtained automatically, will have great utility in future studies and clinical care for patients affected by autosomal dominant polycystic kidney disease. New image-derived biomarkers for patients affected by autosomal dominant polycystic kidney disease are needed to improve current clinical management. The measurement of total kidney volume (TKV) provides critical information for clinicians to drive care decisions. However, patients with similar TKV may present with very different phenotypes, often requiring subjective decisions based on other factors (e.g., appearance of healthy kidney parenchyma, a few cysts contributing significantly to overall TKV, etc.). In this study, we describe a new technique to individually segment cysts and quantify biometric parameters including cyst volume, cyst number, parenchyma volume, and cyst parenchyma surface area. Using data from the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) study the utility of these new parameters was explored, both quantitatively as well as visually. Total cyst number and cyst parenchyma surface area showed superior prediction of the slope of estimated glomerular filtration rate decline, kidney failure and chronic kidney disease stages 3A, 3B, and 4, compared to TKV. In addition, presentations such as a few large cysts contributing significantly to overall kidney volume were shown to be much better stratified in terms of outcome predictions. Thus, these new image biomarkers, which can be obtained automatically, will have great utility in future studies and clinical care for patients affected by autosomal dominant polycystic kidney disease. Translational StatementIn this study, we evaluated new image-derived biomarkers as prognostic factors of autosomal dominant polycystic kidney disease. We used a deep-learning model to generate instance-level cyst segmentations, a previously unattainable task, that facilitated the characterization of architectural disease presentations beyond total kidney volume (TKV). The new biomarkers include total cyst volume, renal parenchyma volume, total cyst number (TCN), and cyst-parenchyma surface area (CPSA). TCN and CPSA showed improved prediction of disease progression after an 8-year and a 20-year period compared with height-adjusted TKV, an important finding that could enhance existing prognostic tools. In this study, we evaluated new image-derived biomarkers as prognostic factors of autosomal dominant polycystic kidney disease. We used a deep-learning model to generate instance-level cyst segmentations, a previously unattainable task, that facilitated the characterization of architectural disease presentations beyond total kidney volume (TKV). The new biomarkers include total cyst volume, renal parenchyma volume, total cyst number (TCN), and cyst-parenchyma surface area (CPSA). TCN and CPSA showed improved prediction of disease progression after an 8-year and a 20-year period compared with height-adjusted TKV, an important finding that could enhance existing prognostic tools. Autosomal dominant polycystic kidney disease (ADPKD) is the most common monogenic kidney disease characterized by the development and growth of cysts in the kidneys. The rate of ADPKD progression has been shown to vary widely among patients; thus, parameters indicative of disease progression are highly valuable for clinical decision-making. Currently, total kidney volume (TKV) is the main imaging biomarker used for clinical management, outcome prediction, and assessment of the efficacy of novel therapeutics.1Grantham J.J. Torres V.E. Chapman A.B. et al.Volume progression in polycystic kidney disease.N Engl J Med. 2006; 354: 2122-2130Crossref PubMed Scopus (645) Google Scholar, 2Torres V.E. Chapman A.B. Devuyst O. et al.Tolvaptan in patients with autosomal dominant polycystic kidney disease.N Engl J Med. 2012; 367: 2407-2418Crossref PubMed Scopus (1145) Google Scholar, 3Caroli A. Perico N. Perna A. et al.Effect of longacting somatostatin analogue on kidney and cyst growth in autosomal dominant polycystic kidney disease (ALADIN): a randomised, placebo-controlled, multicentre trial.Lancet. 2013; 382: 1485-1495Abstract Full Text Full Text PDF PubMed Scopus (205) Google Scholar, 4Wallace D.P. Hou Y.P. Huang Z.L. et al.Tracking kidney volume in mice with polycystic kidney disease by magnetic resonance imaging.Kidney Int. 2008; 73: 778-781Abstract Full Text Full Text PDF PubMed Scopus (35) Google Scholar, 5Grantham J.J. Torres V.E. The importance of total kidney volume in evaluating progression of polycystic kidney disease.Nat Rev Nephrol. 2016; 12: 667-677Crossref PubMed Scopus (90) Google Scholar, 6Higashihara E. Torres V.E. Chapman A.B. et al.Tolvaptan in autosomal dominant polycystic kidney disease: three years' experience.Clin J Am Soc Nephrol. 2011; 6: 2499-2507Crossref PubMed Scopus (134) Google Scholar, 7Irazabal M.V. Rangel L.J. Bergstralh E.J. et al.Imaging classification of autosomal dominant polycystic kidney disease: a simple model for selecting patients for clinical trials.J Am Soc Nephrol. 2015; 26: 160-172Crossref PubMed Scopus (404) Google Scholar Noninvasive visualizations of the kidneys can be made by various imaging modalities, such as ultrasound, computed tomography, and magnetic resonance (MR). Once the images of the kidneys are acquired, the most common approaches to measure TKV have been the ellipsoid formula,8Higashihara E. Nutahara K. Okegawa T. et al.Kidney volume estimations with ellipsoid equations by magnetic resonance imaging in autosomal dominant polycystic kidney disease.Nephron. 2015; 129: 253-262Crossref PubMed Scopus (30) Google Scholar stereology,9Bae K.T. Tao C. Zhu F. et al.MRI-based kidney volume measurements in ADPKD: reliability and effect of gadolinium enhancement.Clin J Am Soc Nephrol. 2009; 4: 719-725Crossref PubMed Scopus (56) Google Scholar and planimetry. The first 2 methods have been shown to have the lowest accuracy but are faster to measure. Planimetry, on the other hand, has been shown to have the highest accuracy and precision but is more time-consuming to perform manually.10Sharma K. Caroli A. Quach L.V. et al.Kidney volume measurement methods for clinical studies on autosomal dominant polycystic kidney disease.PLoS One. 2017; 12e0178488Crossref Scopus (31) Google Scholar This constraint is no longer a limitation with the development of semiautomated and artificial intelligence–powered (fully automated) segmentation algorithms.11Kline T.L. Korfiatis P. Edwards M.E. et al.Performance of an artificial multi-observer deep neural network for fully automated segmentation of polycystic kidneys.J Digit Imaging. 2017; 30: 442-448Crossref PubMed Scopus (102) Google Scholar,12van Gastel M.D. Edwards M.E. Torres V.E. et al.Automatic measurement of kidney and liver volumes from MR images of patients affected by autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2019; 30: 1514-1522Crossref PubMed Scopus (57) Google Scholar TKV is the only Food and Drug Administration–approved imaging biomarker to be used as a surrogate for kidney function in studies evaluating the efficacy of treatment interventions for patients affected by ADPKD. However, the wide range of phenotypic variations of the disease limit the usefulness of TKV.13Hateboer N. v Dijk M.A. Bogdanova N. et al.Comparison of phenotypes of polycystic kidney disease types 1 and 2.Lancet. 1999; 353: 103-107Abstract Full Text Full Text PDF PubMed Scopus (502) Google Scholar,14Senum S.R. Li Y.S.M. Benson K.A. et al.Monoallelic IFT140 pathogenic variants are an important cause of the autosomal dominant polycystic kidney-spectrum phenotype.Am J Hum Genet. 2022; 109: 136-156Abstract Full Text Full Text PDF PubMed Scopus (43) Google Scholar This has led to studies investigating new imaging biomarkers, including the evaluation of cyst volume through the application of image processing–based intensity threshold methods. Cyst volume was found to increase at a steady rate with close correlation to TKV.1Grantham J.J. Torres V.E. Chapman A.B. et al.Volume progression in polycystic kidney disease.N Engl J Med. 2006; 354: 2122-2130Crossref PubMed Scopus (645) Google Scholar The estimates of cyst number from a few MR slices have been studied and suggest that patients with a lower number of cysts have a milder ADPKD presentation than patients with a higher number of cysts.15Bae K.T. Zhou W. Shen C. et al.Growth pattern of kidney cyst number and volume in autosomal dominant polycystic kidney disease.Clin J Am Soc Nephrol. 2019; 14: 823-833Crossref PubMed Scopus (21) Google Scholar,16Harris P.C. Bae K.T. Rossetti S. et al.Cyst number but not the rate of cystic growth is associated with the mutated gene in autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2006; 17: 3013-3019Crossref PubMed Scopus (200) Google Scholar Other efforts relate to the development of new image acquisition protocols for the characterization of renal tissue using magnetization transfer imaging17Kline T.L. Irazabal M.V. Ebrahimi B. et al.Utilizing magnetization transfer imaging to investigate tissue remodeling in a murine model of autosomal dominant polycystic kidney disease.Magn Reson Med. 2016; 75: 1466-1473Crossref PubMed Scopus (32) Google Scholar and image-processing techniques to extract textural features18Kline T.L. Korfiatis P. Edwards M.E. et al.Image texture features predict renal function decline in patients with autosomal dominant polycystic kidney disease.Kidney Int. 2017; 92: 1206-1216Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar; however, these methods are challenging to apply to retrospective data. We recently developed a novel automated approach that facilitates the segmentation of individual cysts,19Gregory A.V. Anaam D.A. Vercnocke A.J. et al.Semantic instance segmentation of kidney cysts in MR images: a fully automated 3D approach developed through active learning.J Digit Imaging. 2021; 34: 773-787Crossref PubMed Scopus (13) Google Scholar an impractical and time-consuming task for humans to perform. This has opened the opportunity to explore more precise and new image-derived biomarkers. These include measurements of total cyst volume (TCV) defined as the sum of all cyst volumes, renal parenchyma volume (RPV) defined as the difference between TKV and TCV, total cyst number (TCN) defined as the count of all cysts, and cyst-parenchyma surface area (CPSA) defined as the sum of all cyst surface areas covered by renal parenchyma (i.e., the sum of all cyst surface areas minus the outer surface of exophytic cysts). In this retrospective study, we evaluate the utility of the new biomarkers on the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) dataset20Chapman A.B. Guay-Woodford L.M. Grantham J.J. et al.Renal structure in early autosomal-dominant polycystic kidney disease (ADPKD): the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) cohort.Kidney Int. 2003; 64: 1035-1045Abstract Full Text Full Text PDF PubMed Scopus (330) Google Scholar to understand how these new image-derived measurements can improve the quantification of disease presentations and better stratify patients beyond a simple measurement of TKV. This study was reviewed and approved by our institution’s institutional review board and was HIPAA (Health Insurance Portability and Accountability Act) compliant. A cohort of the CRISP study including 232 ADPKD patients without azotemia and with creatinine clearance >70 ml/min were included in this J.J. Torres V.E. Chapman A.B. et al.Volume progression in polycystic kidney disease.N Engl J Med. 2006; 354: 2122-2130Crossref PubMed Scopus (645) Google Scholar The CRISP were evaluated and for a 20-year MR images and between and were of the and the clinical such as estimated glomerular filtration rate and were for all Kidney volumes obtained by the ellipsoid and methods were for with the approach The Kidney Disease is was used to The imaging biomarkers were evaluated based on the The slope of estimated using for patients with at between and the 8-year was as the The outcome was kidney failure after from were based on and patients without were based on the of measurement at the 20-year a not chronic kidney disease stages 3A, 3B, and 4, and based on the measurements at the 8-year is in the Kidney segmentation was using a previously developed T.L. Korfiatis P. Edwards M.E. et al.Performance of an artificial multi-observer deep neural network for fully automated segmentation of polycystic kidneys.J Digit Imaging. 2017; 30: 442-448Crossref PubMed Scopus (102) Google Scholar,12van Gastel M.D. Edwards M.E. Torres V.E. et al.Automatic measurement of kidney and liver volumes from MR images of patients affected by autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2019; 30: 1514-1522Crossref PubMed Scopus (57) Google Scholar and the was reviewed by a imaging The MR image and kidney segmentation were the cyst segmentation A.V. Anaam D.A. Vercnocke A.J. et al.Semantic instance segmentation of kidney cysts in MR images: a fully automated 3D approach developed through active learning.J Digit Imaging. 2021; 34: 773-787Crossref PubMed Scopus (13) Google Scholar The of cyst individually The a review by 2 imaging to the clinical using an T.L. Edwards M.E. Korfiatis P. et segmentation of polycystic kidneys in MR Am J 2016; PubMed Scopus (31) Google Scholar TCV, and CPSA were using and from the J.J. A. C. et to the 2017; PubMed Scopus Google Scholar The of the are in the and and and were by the volumes with the was as the difference between and A threshold cyst of was to cysts to MR imaging is in The was using the and the The was to evaluate the between TKV methods. was to evaluate imaging to predict the slope of were estimated using the and the area the with were the were using the The was used to measure the prediction of the new biomarkers and TKV methods. MR images from the CRISP study were as shown in images were the segmentation review of imaging large image cyst other in the cysts, TCN images large to (e.g., image the prediction of cysts was CPSA measurements could be a than a large can the of cysts the image slices with higher data at the 8-year were for A total of patients the of the measurements were than ml/min at of the are shown in and clinical slope of and ml/min estimated glomerular filtration kidney polycystic kidney disease. in a new estimated glomerular filtration kidney polycystic kidney disease. In the the and of between the TKV measurements estimated by and the ellipsoid and between and the The in both was than however, of were between and with an of of the imaging biomarkers from the in a of at The was The was The TCN was cysts cysts). The CPSA was The correlation of all imaging biomarkers, and can be found in MR image with parenchyma, and CPSA segmentation and 3D is shown in A the imaging biomarkers with the slope after an 8-year was The imaging biomarkers and CPSA showed a correlation with the 8-year slope of with a correlation of and TCN and CPSA showed the highest correlation with the of the for the 8-year can be found in shown in of patients were to the imaging biomarkers. was and was based on similar but different with a slope decline after are with MR and patients with a slope decline after are with MR The imaging biomarkers, and from the are shown in of the patients shown in cyst-parenchyma surface height-adjusted renal parenchyma height-adjusted total cyst height-adjusted total kidney volume total cyst in a new cyst-parenchyma surface height-adjusted renal parenchyma height-adjusted total cyst height-adjusted total kidney volume total cyst The of the imaging biomarkers for progression to after a 20-year was superior to as shown by the and CPSA a with an than In addition, CPSA a significantly higher compared with in the prediction of of the progression to chronic kidney disease stages 3A, 3B, and and at the 8-year can be found in and for and imaging biomarkers in the progression to kidney failure after a 20-year cyst V.E. Chapman A.B. Devuyst O. et al.Tolvaptan in patients with autosomal dominant polycystic kidney disease.N Engl J Med. 2012; 367: 2407-2418Crossref PubMed Scopus (1145) Google cyst-parenchyma surface height-adjusted renal parenchyma height-adjusted total cyst height-adjusted total kidney volume kidney total cyst data in a new cyst-parenchyma surface height-adjusted renal parenchyma height-adjusted total cyst height-adjusted total kidney volume kidney total cyst data In a the from the methods and and the automated and CPSA in progression to after the 20-year is understand how the individual biomarkers could improve in terms of 20-year patients and we the by methods with the biomarkers obtained of the new imaging biomarkers improved compared with the ellipsoid The was for for for for and for with the was for for for for and for In this study, we the prognostic of novel imaging biomarkers on the CRISP and imaging biomarkers from MR images were to predict the slope of and after a period of and The were using 2 neural network The first model kidney and the model instance-level cyst T.L. Korfiatis P. Edwards M.E. et al.Performance of an artificial multi-observer deep neural network for fully automated segmentation of polycystic kidneys.J Digit Imaging. 2017; 30: 442-448Crossref PubMed Scopus (102) Google Scholar,12van Gastel M.D. Edwards M.E. Torres V.E. et al.Automatic measurement of kidney and liver volumes from MR images of patients affected by autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2019; 30: 1514-1522Crossref PubMed Scopus (57) Google A.V. Anaam D.A. Vercnocke A.J. et al.Semantic instance segmentation of kidney cysts in MR images: a fully automated 3D approach developed through active learning.J Digit Imaging. 2021; 34: 773-787Crossref PubMed Scopus (13) Google Scholar These can generate in a few The of image biomarkers from the images and can be in a few on a In this study, the a by 2 imaging the of the imaging thus, the parameters are not based on deep-learning model the of imaging biomarker was evaluated and compared with the of TKV, the only imaging biomarker used A study evaluating methods for the of TKV has that measurements by are more and however, of the to perform methods are often K. Caroli A. Quach L.V. et al.Kidney volume measurement methods for clinical studies on autosomal dominant polycystic kidney disease.PLoS One. 2017; 12e0178488Crossref Scopus (31) Google Scholar segmentation methods the of TKV by Gastel M.D. Edwards M.E. Torres V.E. et al.Automatic measurement of kidney and liver volumes from MR images of patients affected by autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2019; 30: 1514-1522Crossref PubMed Scopus (57) Google Scholar In our study, among and the ellipsoid and showed the highest The of in that the TKV methods have a similar area the 20-year however, the a higher the ellipsoid to the The new imaging biomarkers and CPSA were found to predict kidney function decline better than in the 8-year as shown in the in was not as but highly with correlation in The imaging biomarkers in patients with and without were and TCN was the imaging biomarker with the lowest correlation with this finding that TCN provides phenotypic information not by TCN features to a evaluation of TCN could information the rate of cyst studies using an of the TCN that and have in cyst number and volume but no in the rate of of cyst number K.T. Zhou W. Shen C. et al.Growth pattern of kidney cyst number and volume in autosomal dominant polycystic kidney disease.Clin J Am Soc Nephrol. 2019; 14: 823-833Crossref PubMed Scopus (21) Google Scholar,16Harris P.C. Bae K.T. Rossetti S. et al.Cyst number but not the rate of cystic growth is associated with the mutated gene in autosomal dominant polycystic kidney disease.J Am Soc Nephrol. 2006; 17: 3013-3019Crossref PubMed Scopus (200) Google Scholar cysts may be the MR image and may not be for by the these texture to the of the renal T.L. Korfiatis P. Edwards M.E. et al.Image texture features predict renal function decline in patients with autosomal dominant polycystic kidney disease.Kidney Int. 2017; 92: 1206-1216Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar However, texture features are much more to image acquisition parameters and have for of the CPSA has a and to an individual on kidney function (e.g., exophytic cysts will have a than the cysts by kidney the of showed that patients with kidney function decline with cysts, with kidney function with and cysts and in with exophytic the new imaging biomarkers showed the with the slope of are other and factors could be contributing to the in the rate of disease but the study of these factors is of the of this We that this will as a for future model development that may in better prediction of disease in the has efforts such as could in the of The and model are made at imaging biomarkers as for We that parenchyma volumes with a slope of decline in in total kidney and cyst volume and in parenchyma volume and were in a study of patients with ADPKD using computed Bergstralh E.J. et and of renal and renal parenchyma volumes in autosomal dominant polycystic kidney disease.J Am Soc Nephrol. PubMed Google Scholar of in the CRISP with kidney function at the study, showed a correlation between and the slope of A could be to the of and of by the MR image is that the CRISP study of the of in patients with ADPKD a period of glomerular in the patients with the most S. Shen C. D.P. et of kidney function in autosomal-dominant polycystic kidney disease.Kidney Int. 2019; Full Text Full Text PDF PubMed Scopus Google Scholar to the of the liver in polycystic liver K. Torres V.E. et in early autosomal-dominant polycystic kidney disease.Clin 2015; Full Text Full Text PDF PubMed Scopus Google Scholar could be the development of The parenchyma has been previously to as et evaluation of autosomal dominant polycystic kidney disease a J Am Soc Nephrol. 2006; PubMed Scopus (35) Google Scholar The study by Caroli et A. S. et volume on computed imaging a that glomerular filtration rate decline in autosomal dominant polycystic kidney disease J 2011; Full Text Full Text PDF PubMed Scopus Google Scholar the volume by an intensity in computed and showed that the of volume better with that the computed volume to of from A. S. et volume on computed imaging a that glomerular filtration rate decline in autosomal dominant polycystic kidney disease J 2011; Full Text Full Text PDF PubMed Scopus Google Scholar In MR T.L. Irazabal M.V. Ebrahimi B. et al.Utilizing magnetization transfer imaging to investigate tissue remodeling in a murine model of autosomal dominant polycystic kidney disease.Magn Reson Med. 2016; 75: 1466-1473Crossref PubMed Scopus (32) Google T.L. Edwards M.E. et of kidneys in renal PubMed Scopus Google Scholar texture T.L. Korfiatis P. Edwards M.E. et al.Image texture features predict renal function decline in patients with autosomal dominant polycystic kidney disease.Kidney Int. 2017; 92: 1206-1216Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar may from parenchyma to the of is to the of volume as the to faster of the of this study include the for MR images to imaging biomarkers may be more than the for of patients were of MR image The of imaging can the quantification of imaging biomarkers. included the of cyst on slices that affected the of cysts and image acquisition the accuracy of surface area large the of cysts from to and could limit the of A limitation was the of patients between the cyst instance-level model development and the CRISP the images used for model were at different compared with the CRISP all the cyst instance-level were for by imaging the of imaging biomarkers. In the new cyst imaging biomarkers can information to the presentation of ADPKD beyond the information by TKV. This new information has the to improve current of disease progression and to a more model of care the phenotypic between patients affected by ADPKD. Studies with are to and new to investigate the of the new imaging biomarkers with the no of this data be a of HIPAA and the by our institutional review board to this This was by the of and and Kidney and with of the of the cyst-parenchyma surface cysts that were based on the threshold of between imaging biomarkers and the estimated glomerular filtration rate and at of the 8-year slope of estimated glomerular filtration rate and imaging biomarkers. for and imaging biomarkers in the progression to kidney failure and chronic kidney disease stages 3A, 3B, and after an 8-year the for height-adjusted total kidney volume height-adjusted total cyst volume height-adjusted renal parenchyma volume total cyst number (TCN), and cyst-parenchyma surface area (CPSA) for progression to chronic kidney disease stages 3A, 3B, and 4, and kidney failure after the 8-year the for the prognostic imaging biomarker in kidney volume the most prognostic biomarker for autosomal dominant polycystic kidney cyst growth that kidney function of glomerular filtration rate however, by total kidney volume, and is Using deep et total kidney volume and biomarkers. and improved prognostic accuracy for kidney function the study and for PDF

Topics & Concepts