Kidney International, Vol. 61, Supplement 80 (2002), pp. S121–S125

The bioartificial kidney in the treatment of acute renal failure

H. DAVID HUMES, WILLIAM H. FISSELL, and WILLIAM F. WEITZEL

Departments of Internal Medicine, VA Medical Center and The University of Michigan, Ann Arbor, Michigan, USA

The bioartificial kidney in the treatment of acute renal failure. renal substitution processes of hemodialysis and hemo-The application of cell therapy to the successful substitution filtration [4].process of hemofiltration may improve the poor prognosis of The approach to patients with chronic or acute renalpatients with acute renal failure (ARF) in the intensive care

failure (ARF) with dialysis or hemofiltration is still sub-unit. An extracorporeal bioartificial kidney consisting of a con-optimal in terms of morbidity and mortality [5, 6]. Poorventional hemofilter followed in series with a renal tubule assist

device (RAD) has been developed. The RAD is a hemofiltra- outcomes may be due to the fact that these therapiestion cartridge containing 109 human renal tubule cells grown substitute for the small solute clearance function of theas monolayers along the inner surface of the hollow fibers. The kidney but do not replace the lost reclamation, meta-fibers provide a porous scaffold that is immunoprotective. The

bolic, and endocrine functions of the kidney [7]. The dis-ultrafiltrate from the hemofilter is delivered to the luminalorder of ARF, or acute tubular necrosis (ATN), may becompartment of the RAD, and the postfiltered blood is deliv-especially amenable to cell therapy in conjunction withered to the extracapillary space of the RAD.

The RAD has been shown to possess multiple differentiated continuous hemofiltration techniques, because acute he-transport, metabolic, and endocrinologic activities of renal epi- modialysis or hemofiltration has yet to reduce the mor-thelium. These activities have been demonstrated to occur tality rate of ATN below 50%, despite advances in syn-when the RAD is placed in the extracorporeal circuit of the

thetic materials and extracorporeal circuits.bioartificial kidney in uremic animals. This approach may im-ATN develops predominantly from the injury and ne-prove the current therapies used to treat patients with ARF

because of the RAD’s ability to restore lost metabolic renal crosis of renal proximal tubule cells. Replacement of thefunction and cytokine balance in these desperately ill patients. functions of these cells during the episode of ATN willIn this regard, the RAD was able to ameliorate endotoxin and provide almost full renal replacement therapy in con-bacteremic shock in uremic animals by altering cytokine levels,

junction with hemofiltration. The addition of metabolicimprove mean arterial blood pressure, and maintain betteractivity (ammoniagenesis and glutathione reclamation),cardiac output. With these supportive preclinical data, an FDA-

approved phase I/II clinical trial has been initiated and early endocrine activity (vitamin D3 activation), and cytokineresults are encouraging. homeostasis may provide additional physiologic replace-

ment activities to change the current natural history ofthis disease process [7].

Cell therapy is a new and exciting approach to thetreatment of acute and chronic diseases [1]. The potential

BIOARTIFICIAL RENAL TUBULEsuccess of this therapeutic approach lies in the growingASSIST DEVICEappreciation that most disease processes are not caused

Our laboratory has recently developed an extracorpo-by the lack of a single protein but develop from alter-real device utilizing a standard hemofiltration cartridgeations in complex interactions of a variety of cell prod-containing over 109 renal tubule cells grown as confluentucts. Cell therapy is dependent on cell and tissue culturemonolayers along the inner surface of the fibers [8–10].methodologies to expand specific cells to replace impor-The nonbiodegradability and the pore size of the hollowtant differentiated processes deranged or lost in variousfibers allow the membranes to act as both scaffolds for thedisease states. Recent approaches have made progresscells and as an immunoprotective barrier. In vitro studiesby placing cells into hollow fiber bioreactors or encap-of this renal tubule assist device (RAD) have demon-sulating membranes as a means to deliver cell activitiesstrated that the cells retain differentiated active transportto a patient [2, 3]. A reasonable extension of this ap-properties, differentiated metabolic activities, and impor-proach is to add cell therapy to the current successfultant endocrine processes. Additional studies have shownthat the RAD, when incorporated in series with a hemo-

Key words: kidney tubule cells, cell therapy, acute tubular necrosis. filtration cartridge in an extracorporeal blood perfusioncircuit, replaces filtration, transport, and metabolic and 2002 by the International Society of Nephrology

S-121

Humes et al: Bioartificial kidney and acute renal failureS-122

endocrine functions of the kidney in acutely uremic dogs applicable device with the use of commercially availablehigh-flux hemofiltration hollow fiber cartridges. Experi-[4]. Furthermore, recent studies from our laboratory

have demonstrated that the RAD ameliorates endotoxin ments have tested transport and metabolic functions ofthe cells grown intraluminally within these cartridges withshock in acutely uremic animals [11].

The previous renal tubule cell constructs utilized por- membrane surface areas of 97 cm2 to 0.7 m2 [20]. Startingwith a hemofilter cartridge, the intraluminal surface ofcine cells because the pig was initially considered the

best source of organs for both xenotransplantation and the hollow fibers was coated with an extracellular matrixmolecule. Renal tubule cells were then seeded at a den-cell therapy devices because of the anatomical and physi-

ologic similarities to human tissue and the relative ease sity of 105 cells/mL into the intracapillary space. The seededcartridge was connected to the bioreactor perfusion sys-of breeding large numbers of pigs in closed herds [12].

Recent reports of the ability of porcine endogenous ret- tem, in which the extracapillary space was filled withculture media and the intracapillary space perfused withroviruses (PERVs) to infect human cells in co-culture

in vitro, however, have raised concerns of the potential, similar media. After 7 to 14 days of growth, the unit wasstudied. In vitro experiments utilizing porcine renal prox-but currently unquantifiable, risk of transmission of viral

elements between species with the use of porcine tissue imal tubule progenitor cells have clearly shown differen-tiated transport and metabolic functions of the RADin xenotransplantation or cell therapy devices [13, 14].

Accordingly, we have successfully demonstrated meth- unit [20].The bioartificial kidney setup consists of a filtrationodologies to fabricate RADs with human renal tubule

cells [15] and have begun phase I/II trials with this device device (a conventional hemofilter) followed in series bythe tubule RAD unit [4]. Specifically, blood is pumpedin patients with ARF in the intensive care unit.

Critical to providing organ function replacement through out of a large animal using a peristaltic pump. The bloodthen enters the fibers of a hemofilter, where ultrafiltratecell therapy is the need for the isolation and growth in

vitro of specific cells from adult tissue. These cells possess is formed and delivered into the fibers of the tubulelumens within the RAD downstream to the hemofilter.stem cell-like characteristics with a high capacity for self-

renewal and the ability to differentiate under defined con- Processed ultrafiltrate exiting the RAD is collected anddiscarded as “urine.” The filtered blood exiting the hemo-ditions into specialized cells to develop correct structure

and functional components of a physiologic organ system filter enters the RAD through the extracapillary spaceport and disperses among the fibers of the device. Upon[16, 17]. Data from our laboratory have shown a method-

ology to isolate and grow renal proximal tubule progeni- exiting the RAD, the processed blood travels through athird pump and is delivered back to the animal. Thistor cells from adult mammalian kidneys [18, 19]. These

studies were promoted by the clinical and experimental additional pump is required to maintain appropriate hy-draulic pressures within the RAD. In this regard, theobservations suggesting that renal proximal tubule pro-

genitor cells must exist, as tubule cells have the ability pressures of the blood and ultrafiltrate just before entryinto the RAD are monitored. Heparin is delivered con-to regenerate after severe nephrotoxic or ischemic injury

to form a fully functional and differentiated epithelium. tinuously into the blood before entering the RAD todiminish clotting within the device. The RAD is orientedThe bioartificial renal tubule clearly is feasible when

conceived as a combination of living cells supported on horizontally and placed into a temperature-controlledenvironment. The temperature of the cell compartmentpolymeric substrata. A bioartificial tubule uses epithelial

progenitor cells cultured on biomatrix-coated hollow fi- of the RAD must be maintained at 37�C throughout itsoperation to ensure optimal functionality of the cells.ber membranes that are both water- and solute-perme-

able, allowing for the differentiated vectorial transport Maintenance of a physiologic temperature is a criticalfactor in the functionality of the RAD.and metabolic and endocrine function. With appropriate

membranes and biomatrices, immunoprotection of cul- The tubule unit is able to maintain viability becausemetabolic substrates and low-molecular weight growthtured progenitor cells can be achieved concurrently with

long-term functional performance as long as conditions factors are delivered to the tubule cells from the ultrafil-tration unit and the blood in the extracapillary space.support tubule cell viability [8]. The technical feasibility

of an epithelial cell system derived from cells grown as Furthermore, immunoprotection of the cells grown withinthe hollow fiber is achieved because of the impenetrabil-confluent monolayers along the luminal surface of a sin-

gle polymeric hollow fiber has been achieved [9]. ity of immunoglobulins and immunologically competentcells through the hollow fibers. Rejection of the cells,The next step in the development of a RAD was to

scale up from this single hollow fiber renal tubule to a therefore, does not occur. This arrangement thereby al-lows the filtrate to enter the internal compartments of themultifiber bioreactor with renal proximal tubule cells that

maintain not only transport properties but also differen- hollow fiber network, lined with confluent monolayers ofrenal tubule cells for regulated transport and metabolictiated metabolic and endocrine functions. Completed

experiments have successfully scaled up to a clinically function.

Humes et al: Bioartificial kidney and acute renal failure S-123

BIOARTIFICIAL KIDNEY IN adequate nutrition, and treatment of infection and ure-mia when they are present. Uremia is treated with eitherUREMIC ANIMALSintermittent hemodialysis or continuous hemofiltration.Large animal studies have been completed with the useAlthough this approach has had substantial impact onof this extracorporeal circuit [4]. Dogs were made uremicthis disease process over the past 40 years, patients withby performing bilateral nephrectomies. A double lumenATN still have an exceedingly high mortality of greatercatheter was placed into the internal jugular vein, ex-than 50%, even with dialytic or hemofiltrative supporttending into the heart. After 24 hours of postoperative[21, 22]. Mortality remains high despite normal electro-recovery, the dogs were treated either with hemofiltra-lyte balance and improvement in the uremic state causedtion and the RAD or with hemofiltration and a shamby the propensity of these patients to develop a systemiccontrol cartridge containing no cells. The blood flow rateinflammatory response syndrome (SIRS), most com-to the hemofiltrator was maintained at 80 mL/min, withmonly secondary to bacterial sepsis, with resulting multi-a controlled ultrafiltration rate of 5 to 7 mL/min. Dogsorgan damage from cardiovascular collapse and ischemiawere treated daily for either 7 or 9 hours on 3 successive[23].days or for 24 hours continuously.

SIRS is a catastrophic sequela of a variety of clinicalThe dogs in these experiments developed ARF withinsults, including infection, pancreatitis, and cardiopul-average BUN and plasma creatinine levels of 68 and 6.6monary bypass, and claims over a quarter million livesmg/dL, respectively. The RADs maintained viability andin the United States each year [24–26]. The exceptionallyfunctionality when connected in series to a hemofiltra-high mortality associated with the syndrome is causedtion cartridge within an extracorporeal perfusion circuitin part by the development of the highly lethal multiplein an acutely uremic animal. During a 24-hour perfusionsystem organ failure syndrome (MSOF) in a subset ofperiod, fewer than 105 cells were lost from the RAD,patients with SIRS. The sequential failure of organ sys-which contained more than 109 cells. Treatment with thetems apparently unrelated to the site of the initial insultRAD and hemofiltration maintained BUN and plasmahas been correlated with altered plasma cytokine levelscreatinine levels similar to those of sham controls. Inobserved in sepsis [27–30]. The mortality is especiallyaddition, plasma HCO�

3 , Pi, and K� levels were morehigh in patients with MSOF and ARF. The excess mortal-readily maintained near normal values in RAD treat-ity seen in patients with sepsis and ARF is not amelio-ment than in sham treatment. The RADs were able torated by conventional renal replacement therapy, whichreabsorb 40–50% of ultrafiltrate volume presented to thetreats volume overload, uremia, acidosis, and electrolytedevices. Furthermore, active transport of K�, HCO�

3 ,derangements [31].and glucose was accomplished by the RAD in this ex

The RAD may alter the natural history of this disordervivo situation. Metabolic activity of the RAD was alsoif the renal tubular epithelium has an immunoregulatoryshown in these experiments. Virtually no ammonia ex-role in whole body homeostasis. In this regard, it has beencretion occurred in the processed ultrafiltrate of the shamnoted that patients with chronic renal failure are morecontrol group, in contrast with an ammonia excretionsusceptible to infection despite adequate dialysis [32].level as high as 100 �mol/h in the RAD-treated group.This observation, as well as the correlation of postvacci-Glutathione processing by the RAD was also shown,nation immunity with cytokine levels in renal failurewith greater than 50% glutathione removal from the ultra-patients, also suggests an immunologic role for the kid-filtrate presented to the RAD. Finally, uremic animalsney [33, 34]. There are several plausible mechanisms bytreated with the RAD attained 1,25-(OH)2-D3 levels ofwhich this could occur, among them immune surveillance19.5 � 0.5 pg/mL, a value no different from the normalby antigen-presenting cells in the kidney, antioxidant sup-levels of the prenephrectomy condition. In contrast, shamply by glutathione synthesis and glutathione peroxidasetreatment resulted in a further fall of 4.0 � 2.4 pg/mLin the proximal tubule cell, and cytokine regulation byfrom the already low plasma levels of 1,25-(OH)2-D3tubule cells. In addition, low 1,25-(OH)2-D3 levels in hos-in the acutely uremic animals. Thus, these experimentspitalized patients appear to have a correlation with mor-clearly showed that the combination of a synthetic hemo-tality rates in these patients [35], suggesting yet anotherfiltration cartridge and a RAD in an extracorporeal cir-influence of the kidney on immune competence. Our lab-cuit successfully replaced filtration, transport, and meta-oratory recently devised a series of experiments to testbolic and endocrinologic functions of the kidney inthis hypothesis.acutely uremic dogs.

These experiments investigated whether treatment witha bioartificial kidney would alter the course of sepsis in

POTENTIAL IMPACT OF RAD IN ARF an animal model. [11] Mongrel dogs (N � 10) underwentsurgical nephrectomy and 48 hours later were treatedCurrent therapy for ischemic or toxic ARF, or ATN, is

predominantly supportive in nature. The therapeutic goals with continuous venovenous hemofiltration (CVVH)and either a RAD containing cells (N � 5) or an identi-are the maintenance of fluid and electrolyte balance,

Humes et al: Bioartificial kidney and acute renal failureS-124

cally prepared sham cartridge (N � 5). After four hours with CVVH (abstract; Weitzel et al, J Am Soc Nephrolof therapy, 2 mg/kg intravenous endotoxin was infused 12:279A, 2001). These data also indicate that this deviceover one hour to simulate gram-negative septic shock. maintains and exhibits viability, durability, and function-Data on blood pressure, cardiac output, and systemic ality in this ex vivo clinical setting. The cardiovascularmarkers of inflammation were collected. Mean peak lev- stability of the patients was maintained with RAD treat-els of an anti-inflammatory cytokine, IL-10, were signifi- ment in two of the three patients and was manageablecantly higher in cell-treated animals (15.25 ng/mL vs. with pressor medications and fluids in the third patient.6.29 ng/mL; P � 0.037), and mean arterial pressures were The data are presently too few to determine if the bene-higher in cell-treated versus sham-treated animals (P � ficial hemodynamics seen in prior large animal studies0.04). These data demonstrated that treatment of an can be reproduced in humans. More data are requiredanimal model of endotoxin shock and renal failure with to determine if the postulated changes in metabolic re-a bioartificial kidney has measureable effects on circulat- placement translate into improved hemodynamics oring mediators of inflammation and on hemodynamic sta- other clinical benefits. The increased native renal func-bility of the challenged animal. tion temporally correlated with RAD treatment, most

To further assess the role of the bioartificial kidney pronounced in one of the three patients, requires addi-in sepsis in acute renal failure, dogs were nephrectomized tional study. The isolated and expanded human cells alsoand 48 hours later administered intraperitoneally with demonstrated differentiated metabolic and endocrino-30 � 1010 bacteria/kg of E. coli (abstract; Fissell et al, logic activity in this ex vivo treatment. RAD tubule cellJ Am Soc Nephrol 12:264A, 2001). One hour after bacte- GSH degradation activity was maintained, and endocri-rial administration, animals were placed in a CVVH cir- nologic conversion of 25-OH-D3 to 1,25-(OH)2-D3 wascuit with either a sham RAD without cells or a RAD

demonstrated. Changes in serum cytokines are currentlywith cells. Blood pressure, cardiac output, heart rate,

under evaluation, since previous results showed alter-pulmonary capillary wedge pressure, and systemic vascu-ations in serum levels that correlated with improved he-lar resistance were measured throughout the study. Allmodynamics [11].animals tested were in renal failure with BUN and serum

These results provide important early observations increatinine concentrations greater than 60 and 6 mg/dL,the evaluation of the human RAD in the clinical setting.respectively. RAD treatment maintained better cardio-They demonstrate that renal tubule cells can be isolatedvascular performance, as determined by mean arterialfrom human kidneys, expanded and seeded into a hemo-blood pressure and cardiac output, for longer periodsfiltration cartridge to produce a device containing overthan sham RAD therapy. Consistently, all sham RAD109 cells suitable for clinical study. Human cell RADanimals expired within 2–10 hours after bacterial admin-cartridges maintained viability in an extracorporeal he-istration, whereas all RAD-treated animals survivedmoperfusion circuit in series with a synthetic hemofilterlonger than 10 hours.during CVVH in the clinical setting. The durability ofPlasma cytokine levels in the bacteremic animals werethe device was sustained for up to 24 hours of continuousassessed with RAD treatment. The cell RAD maintaineduse, the longest time interval tested in this study. Becauserenal metabolic activity throughout the septic period.the RAD may replace metabolic processes lost in renalTherefore, these data provide further support that cellfailure, further clinical testing is required to assess safetytherapy with the RAD provided metabolic renal replace-and to establish potential clinical efficacy in the acute re-ment resulting in protection against gram-negative sep-nal failure setting. The replacement of renal proximal tu-tic shock.bule cell function with this cell therapy device is a logicaltherapeutic direction, as the pathophysiology of this dis-

INITIAL CLINICAL EXPERIENCE WITH A order is predominantly caused by proximal tubule cellHUMAN RAD injury, apoptosis, and necrosis. Future stages of clinical

With this suggestive preclinical data, the FDA ap- study will focus on adjusting the dose and extending treat-proved an Investigational New Drug application to study ment time while expanding the evaluation of safety andthe RAD containing human cells in patients with acute device function. Ultimately, a controlled phase III efficacytubular necrosis receiving CVVH. Human kidney cells trial will be required to assess whether this new approachwere isolated from kidneys donated for cadaveric trans- can improve the treatment of this disease process, whichplantation but which could not be used for this purpose currently has unacceptably high mortality rates.because of anatomic or fibrotic defects. The initial results

Reprint requests to H. David Humes, M.D., Department of Inter-on the first three treated patients demonstrate that thisnal Medicine, University of Michigan Medical School, 7220 MSRB III,experimental treatment can be delivered safely underBox 0644, 1150 W Medical Center Drive, Ann Arbor, Michigan 48109-

study protocol guidelines in this critically ill patient pop- 0644, USA.E-mail: dhumes@umich.eduulation for up to 24 hours when used in conjunction

Humes et al: Bioartificial kidney and acute renal failure S-125

isolated single cells of adult mammalian kidney: Clonal analysisREFERENCESwith a recombinant retrovirus. Am J Physiol 271:F42–F49, 1996

20. Humes HD, MacKay SM, Funke AJ, et al: Tissue engineering of1. Gage FH: Cell therapy. Nature 392:518–524, 19982. Watanabe FD, Mullon CJ, Hewitt WR, et al: Clinical experience a bioartificial renal tubule assist device: In vitro transport and

metabolic characteristics. Kidney Int 55:2502–2514, 1999with a bioartificial liver in the treatment of severe liver failure: A21. Humes HD, MacKay SM, Funke AJ, et al: The bioartificial renalphase I clinical trial. Ann Surg 225:484–491, 1997

tubule assist device to enhance CRRT in acute renal failure. Am3. Maki T, Otsu I, O’Neil JJ, et al: Treatment of diabetes by xenoge-J Kidney Dis 30(Suppl):S28–S31, 1997neic islets without immunosuppression: Use of a vascularized bioar-

22. Humes HD, MacKay SM, Funke AJ, et al: Acute renal failure:tificial pancreas. Diabetes 45:342–347, 1996Growth factors, cell therapy, gene therapy. Proc Assoc Am Physi-4. Humes HD, Buffington DA, MacKay SM, et al: Replacement ofcians 109:547–557, 1997renal function in uremic animals with a tissue-engineered kidney.

23. Bone RC, Grodzin CJ, Balk RA: Sepsis: A new hypothesis forNat Biotechnol 17:451–455, 1999pathogenesis of the disease process. Chest 112:235–243, 19975. Mehta RL: Therapeutic alternatives to renal replacement for

24. Tran DD, Groenveld AJ, van der Meulen J, et al: Age, chroniccritically ill patients in acute renal failure. Semin Nephrol 14:64–disease, sepsis, organ system failure, and mortality. Crit Care Med82, 199418:474–479, 19906. United States Renal Data System: Annual Data Report. Am J

25. Donnelly S, Robertson C: Mediators, mechanisms and mortalityKidney Dis 32(Suppl):569–580, 1998in major trauma. Resuscitation 28:87–92, 19947. Humes HD: Bioartificial kidney for full renal replacement therapy.

26. Horn K: Evolving strategies in the treatment of sepsis and systemicSemin Nephol 20:71–82, 2000inflammatory response syndrome (SIRS). Quart J Med 91:265–277,8. Nikolovski J, Gulari E, Humes HD: Design engineering of a1998bioartificial renal tubule cell therapy device. Cell Transplant 8:351–

27. Pinsky MR, Vincent JL, Deviere J, et al: Serum cytokine levels364, 1999in human septic shock. Chest 103:565–575, 19939. MacKay SM, Funke AJ, Buffington DA, et al: Tissue engineering

28. Marty C, Misset B, Tamion F, et al: Circulating interleukin-8of a bioartificial renal tubule. ASAIO J 44:179–183, 1998concentrations in patients with multiple organ failure of septic and10. Humes HD, MacKay SM, Funke AJ, et al: Tissue engineering of nonseptic origin. Crit Care Med 22:673–679, 1994a bioartificial renal tubule assist device: In vitro transport and 29. Damas P, Reuter A, Gysen P, et al: Tumor necrosis factor andmetabolic characteristics. Kidney Int 55:2502–2514, 1999 interleukin-1 serum levels during severe sepsis in humans. Crit

11. Fissell WH, Dyke DB, Buffington DA, et al: Bioartificial kidney Care Med 17:975–978, 1989alters cytokine response and hemodynamics in endotoxin-chal- 30. Calandra T, Baumgartner JD, Grau GE, et al: The Swiss-Dutchlenged dogs. Blood Purif 20:55–60, 2002 J5 Immunoglobulin Study Group. Prognostic values of tumor ne-

12. Cozzi E, White D: The generation of transgenic pigs as potential crosis factor/cachectin, interleukin-1, interferon- and interferon-organ donors for humans. Nat Med 1:965–966, 1995 in the serum of patients with septic shock. J Infect Dis 161:982–

13. Le Tissier P, Stoye JP, Takeuchi Y, et al: Two sets of human- 987, 1990tropic pig retrovirus. Nature 389:681–682, 1997 31. Breen D, Bihari D: Acute renal failure as a part of multiple organ

14. Paradis K, Langford G, Long Z, et al: Search for cross-species failure: The slippery slope of critical illness. Kidney Int 53(Suppltransmission of porcine endogenous retrovirus in patients treated 66):S25–S33, 1998with living pig tissue. Science 285:1236–1241, 1999 32. Sarnak M, Jaber B: Mortality caused by sepsis in patients with end-

15. Humes HD, Fissell WH, Weitzel WF, et al: Metabolic replace- stage renal disease compared to the general population. Kidney Intment of renal function in uremic animals with a bioartificial kidney 58:1758–1764, 2000containing human cells. Am J Kidney Dis (in press) 33. Girndt M, Kohler H, Schiedhelm-Weick E: Production of in-

16. Hall PA, Watt PM: The generation and maintenance of cellular terleukin-6, tumor necrosis factor alpha and interleukin-10 in vitrodiversity. Development 106:619–633, 1989 correlates with the clinical immune defect in chronic hemodialysis

17. Potten CS, Loeffler M: Stem cells: Lessons for and from the patients. Kidney Int 47:559–565, 1995crypt. Development 110:1001–1020, 1990 34. Girndt M, Sester U, Sester M, et al: Impaired cellular immune

18. Humes HD, Cieslinski DA: Interaction between growth factors function in patients with end-stage renal failure. Nephrol Dialand retinoic acid in the induction of kidney tubulogenesis. Exp Transplant 14:2807–2830, 1999Cell Res 201:8–15, 1992 35. Thomas M, Lloyd-Jones D, Thadhani R, et al: Hypovitaminosis

19. Humes HD, Krauss JC, Cieslinski DA, et al: Tubulogenesis from in medical inpatients. N Engl J Med 338:777–783, 1998