Mesenchymal stem cells and inflammatory cardiomyopathy: cardiac homing and beyond.

Under conventional heart failure therapy, inflammatory cardiomyopathy usually has a progressive course, merging for alternative interventional strategies. There is accumulating support for the application of cellular transplantation as a strategy to improve myocardial function. Mesenchymal stem cells (MSCs) have the advantage over other stem cells that they possess immunomodulatory features, making them attractive candidates for the treatment of inflammatory cardiomyopathy. Studies in experimental models of inflammatory cardiomyopathy have consistently demonstrated the potential of MSCs to reduce cardiac injury and to improve cardiac function. This paper gives an overview about how inflammation triggers the functionality of MSCs and how it induces cardiac homing. Finally, the potential of intravenous application of MSCs by inflammatory cardiomyopathy is discussed.

Mesenchymal stem cells improve murine acute coxsackievirus B3-induced myocarditis.

AIMS: Coxsackievirus B3 (CVB3)-induced myocarditis, initially considered a sole immune-mediated disease, also results from a direct CVB3-mediated injury of the cardiomyocytes. Mesenchymal stem cells (MSCs) have, besides immunomodulatory, also anti-apoptotic features. In view of clinical translation, we first analysed whether MSCs can be infected by CVB3. Next, we explored whether and how MSCs could reduce the direct CVB3-mediated cardiomyocyte injury and viral progeny release, in vitro, in the absence of immune cells. Finally, we investigated whether MSC application could improve murine acute CVB3-induced myocarditis. METHODS AND RESULTS: Phase contrast pictures and MTS viability assay demonstrated that MSCs did not suffer from CVB3 infection 4-12-24-48 h after CVB3 infection. Coxsackievirus B3 RNA copy number decreased in this time frame, suggesting that no CVB3 replication took place. Co-culture of MSCs with CVB3-infected HL-1 cardiomyocytes resulted in a reduction of CVB3-induced HL-1 apoptosis, oxidative stress, intracellular viral particle production, and viral progeny release in a nitric oxide (NO)-dependent manner. Moreover, MSCs required priming via interferon-γ (IFN-γ) to exert their protective effects. In vivo, MSC application improved the contractility and relaxation parameters in CVB3-induced myocarditis, which was paralleled with a reduction in cardiac apoptosis, cardiomyocyte damage, left ventricular tumour necrosis factor-α mRNA expression, and cardiac mononuclear cell activation. Mesenchymal stem cells reduced the CVB3-induced CD4- and CD8- T cell activation in an NO-dependent way and required IFN-γ priming. CONCLUSION: We conclude that MSCs improve murine acute CVB3-induced myocarditis via their anti-apoptotic and immunomodulatory properties, which occur in an NO-dependent manner and require priming via IFN-γ.

The impact of antigen expression in antigen-presenting cells on humoral immune responses against the transgene product.

Treatment of genetic diseases by gene therapy is hampered by immune responses against the transgene product. Promoter choice has been shown to be an important parameter of the presence or absence of antibodies against the transgene product after gene transfer. Here, the generality of some of these observations was tested by comparing different murine strains and different transgene products. We show immunological unresponsiveness for human apolipoprotein (apo) A-I in six murine strains after transfer with E1E3E4-deleted adenoviral vectors containing hepatocyte-specific expression cassettes. However, differences in the induction of a humoral immune response against human apo A-I after gene transfer with vectors driven by the major histocompatibility complex class II Ebeta promoter and the ubiquitously active cytomegalovirus promoter were not consistent in these six murine strains. Furthermore, use of a potent hepatocyte-specific expression cassette did not prevent a humoral immune response against human plasminogen in C57BL/6 mice. In contrast, human microplasminogen transfer resulted in stable expression in the absence of an immune response against the transgene product. Taken together, the molecular design of strategies to abrogate or induce an immune response against the transgene product may be hampered by the multitude of parameters affecting the outcome, thus limiting the external validity of results.

Enhancement of endothelial nitric oxide synthase production reverses vascular dysfunction and inflammation in the hindlimbs of a rat model of diabetes.

AIMS/HYPOTHESIS: Reduced bioavailability of nitric oxide (NO) is a hallmark of diabetes mellitus-induced vascular complications. In the present study we investigated whether a pharmacological increase of endothelial NO synthase (eNOS) production can restore the impaired hindlimb flow in a rat model of severe diabetes. METHODS: A model of diabetes mellitus was induced in male Sprague-Dawley rats by a single injection of streptozotozin. Rats were treated chronically with the eNOS transcription enhancer AVE3085 (10 mg [kg body weight](-1) day(-1); p.o.) or vehicle for 48 days and compared with controls. Endothelial function and arterial BP were investigated in vivo using an autoperfused hindlimb model and TIP-catheter measurement, respectively. Protein production of eNOS, total and phosphorylated vasodilator-stimulated phosphoprotein (VASP) were assessed in their quadriceps muscle tissue, whereas cyclic GMP (cGMP) concentrations were assessed in blood plasma. RNA levels of intracellular and vascular cell adhesion molecules (ICAM-1 and VCAM-1) were measured by real-time PCR. RESULTS: Untreated diabetic rats showed significantly reduced quadriceps muscle contents of eNOS (-64%) and phosphorylated VASP (-26%) protein associated with impaired vascular function (maximum vasodilatation: -30%, p < 0.05) and enhanced production of ICAM-1 (+121%) and VCAM-1 (+156%). Chronic treatment with AVE3085 did not alter arterial BP or severe hyperglycaemia, but did lead to significantly increased production of eNOS (+95%), cGMP (+128%) and VASP phosphorylation (+65%) as well as to improved vascular function (+36%) associated with reduced production of ICAM-1 (-36%) and VCAM-1 (-58%). CONCLUSIONS/INTERPRETATION: In a rat model of severe diabetes, pharmacological enhancement of impaired eNOS production and NO-cGMP signalling by AVE3085 restores altered hindlimb blood flow and prevents vascular inflammation.

Anti-inflammatory effects of atorvastatin improve left ventricular function in experimental diabetic cardiomyopathy.

AIMS/HYPOTHESIS: Emerging evidence suggests that statins exert beneficial effects beyond those predicted by their cholesterol-lowering actions. We investigated whether atorvastatin influences the development of left ventricular (LV) dysfunction, independently of cholesterol-lowering, in an experimental model of type 1 diabetes mellitus cardiomyopathy. METHODS: Streptozotocin-induced diabetic rats were treated with atorvastatin (50 mg/kg daily, orally) or with vehicle for 6 weeks. LV function was analysed using tip-catheter measurements. Cardiac stainings of TNF-alpha, IL-1beta, intercellular adhesion molecule-1, vascular cellular adhesion molecule-1, CD11a/lymphocyte-associated antigen-1, CD11b/macrophage antigen alpha, CD18/beta2-integrin, ED1/CD68, collagen I and III, and Sirius Red were assessed by digital image analysis. Ras-related C3 botulinum toxin substrate (RAC1) and ras homologue gene family, member A (RHOA) activities were determined by RAC1 glutathione-S-transferase-p21-activated kinase and rhotekin pull-down assays, respectively. Cardiac lipid peroxides were measured by a colorimetric assay. The phosphorylation state of p38 mitogen-activated protein kinase (MAPK) and endothelial nitric oxide synthase (eNOS) protein production were analysed by western blot. RESULTS: Diabetes was associated with induced cardiac stainings of TNF-alpha, IL-1beta, cellular adhesion molecules, increased leucocyte infiltration, macrophage residence and cardiac collagen content. In contrast, atorvastatin reduced both intramyocardial inflammation and myocardial fibrosis, resulting in improved LV function. This effect was paralleled with a normalisation of diabetes-induced RAC1 and RHOA activity, in the absence of LDL-cholesterol lowering. In addition, atorvastatin decreased diabetes-induced cardiac lipid peroxide levels and p38 MAPK phosphorylation by 1.3-fold (p < 0.05) and 3.2-fold (p < 0.0005), respectively, and normalised the reduced eNOS production caused by diabetes. CONCLUSIONS/INTERPRETATION: These data indicate that atorvastatin, independently of its LDL-cholesterol-lowering capacity, reduces intramyocardial inflammation and myocardial fibrosis, resulting in improved LV function in an experimental model of diabetic cardiomyopathy.

Inhibition of p38 mitogen-activated protein kinase attenuates left ventricular dysfunction by mediating pro-inflammatory cardiac cytokine levels in a mouse model of diabetes mellitus.

AIMS/HYPOTHESIS: We investigated the effect of SB 203580, a pharmacological inhibitor of p38 mitogen-activated protein kinase (MAPK), on cardiac inflammation, cardiac fibrosis, and left ventricular function using an animal model of diabetic cardiomyopathy. MATERIALS AND METHODS: Diabetes mellitus was induced by streptozotocin (50 mg/kg i.p. for 5 days) in 20 C57/BL6J mice. Diabetic mice were treated daily with the p38 MAPK inhibitor SB 203580 (1 mg/kg daily, n=10) or with placebo (n=10) and were compared to non-diabetic controls. Left ventricular function was measured by pressure-volume loops after 8 weeks of diabetes mellitus. The parameters for systolic function were the end systolic pressure-volume relationship (ESPVR) and the left ventricular end systolic pressure. The parameters for diastolic function were the left ventricular end diastolic pressure and the end diastolic pressure-volume relationship (EDPVR). Cardiac tissue was analysed by ELISA for the protein content of the cytokines TNF-alpha, IL6, IL1-beta, and TGF-beta1. Phosphorylation of MAPK p38 was analysed by western blot, and the total cardiac collagen content was analysed by Sirius red staining. RESULTS: Left ventricular dysfunction was documented by impaired ESPVR and EDPVR. Cardiac cytokine levels and cardiac fibrosis were increased in diabetic animals compared to controls. Treatment with the p38 inhibitor normalised cardiac cytokine levels and improved systolic function, but did not change cardiac fibrosis and diastolic dysfunction compared to placebo. CONCLUSIONS/INTERPRETATION: Pharmacological inhibition of p38 MAPK prevents cardiac inflammation and attenuates left ventricular dysfunction in diabetic cardiomyopathy.

Benfotiamine accelerates the healing of ischaemic diabetic limbs in mice through protein kinase B/Akt-mediated potentiation of angiogenesis and inhibition of apoptosis.

AIMS/HYPOTHESIS: Benfotiamine, a vitamin B1 analogue, reportedly prevents diabetic microangiopathy. The aim of this study was to evaluate whether benfotiamine is of benefit in reparative neovascularisation using a type I diabetes model of hindlimb ischaemia. We also investigated the involvement of protein kinase B (PKB)/Akt in the therapeutic effects of benfotiamine. METHODS: Streptozotocin-induced diabetic mice, given oral benfotiamine or vehicle, were subjected to unilateral limb ischaemia. Reparative neovascularisation was analysed by histology. The expression of Nos3 and Casp3 was evaluated by real-time PCR, and the activation state of PKB/Akt was assessed by western blot analysis and immunohistochemistry. The functional importance of PKB/Akt in benfotiamine-induced effects was investigated using a dominant-negative construct. RESULTS: Diabetic muscles showed reduced transketolase activity, which was corrected by benfotiamine. Importantly, benfotiamine prevented ischaemia-induced toe necrosis, improved hindlimb perfusion and oxygenation, and restored endothelium-dependent vasodilation. Histological studies revealed the improvement of reparative neovascularisation and the inhibition of endothelial and skeletal muscle cell apoptosis. In addition, benfotiamine prevented the vascular accumulation of advanced glycation end products and the induction of pro-apoptotic caspase-3, while restoring proper expression of Nos3 and Akt in ischaemic muscles. The benefits of benfotiamine were nullified by dominant-negative PKB/Akt. In vitro, benfotiamine stimulated the proliferation of human EPCs, while inhibiting apoptosis induced by high glucose. In diabetic mice, the number of circulating EPCs was reduced, with the deficit being corrected by benfotiamine. CONCLUSIONS/INTERPRETATION: We have demonstrated, for the first time, that benfotiamine aids the post-ischaemic healing of diabetic animals via PKB/Akt-mediated potentiation of angiogenesis and inhibition of apoptosis. In addition, benfotiamine combats the diabetes-induced deficit in endothelial progenitor cells.

Ex vivo gene transfer for improvement of transplanted pancreatic islet viability and function.

Human pancreatic islet transplantation has recently been shown to be successful in replacing pancreatic endocrine function into type 1 diabetic recipients. A major drawback, however, is the high amount of pancreatic ss cells required to render one single patient insulin-independent. Given the shortage of human beta cell donors, the majority of type 1 diabetic patients remain excluded from this therapeutic option. High number of islets are requested since substantial islet cell death and dysfunction occur within the first few hours and days after islet transplantation. Impaired vascularization of the engraft, the non-specific inflammatory reaction at the site of transplantation, together with the presence of active or memory autoimmune responses to islet autoantigens and allogeneic recognition contribute to apoptosis of ss cells and subsequent early graft function loss. This review will focus on ex vivo engineering of the islet graft by gene transfer to improve islet engraftment. An overview of currently used gene transfer techniques will be given and their potential will be discussed.

The size of sinusoidal fenestrae is a critical determinant of hepatocyte transduction after adenoviral gene transfer.

The hepatotropism and intrahepatic distribution of adenoviral vectors may be species dependent. Hepatocyte transduction was evaluated in three rabbit strains after transfer with E1E3E4-deleted adenoviral vectors containing a hepatocyte specific alpha1-antitrypsin promoter-driven expression cassette (AdAT4). Intravenous administration of 4 x 10(12) particles/kg of AdAT4 induced human apo A-I levels above 40 mg/dl in Dutch Belt, but below 1 mg/dl in New Zealand White and Fauve de Bourgogne rabbits. Diameters of sinusoidal fenestrae were significantly (P=0.0014) larger in Dutch Belt (124+/-3.4 nm) than in New Zealand White (108+/-1.3 nm) and Fauve de Bourgogne (105+/-2.6 nm) rabbits, suggesting that a smaller size constitutes a barrier for hepatocyte transduction. Indeed, intraportal transfer preceded by intraportal injection of sodium decanoate, which increases the diameter of sinusoidal fenestrae to 123+/-3.4 nm (P<0.01) in New Zealand White rabbits, increased human apo A-I levels 32- and 120-fold in New Zealand White and Fauve de Bourgogne rabbits, respectively, but did not affect expression in Dutch Belt rabbits. In conclusion, size of sinusoidal fenestrae appears to be a critical determinant of hepatocyte transduction after adenoviral transfer.

Nerve growth factor promotes reparative angiogenesis and inhibits endothelial apoptosis in cutaneous wounds of Type 1 diabetic mice.

AIMS/HYPOTHESIS: The neurotrophin nerve growth factor (NGF) is pro-angiogenic and facilitates wound repair. The present study was conducted to (i) assess the statement of NGF system components in diabetic wounds and (ii) evaluate whether NGF supplementation could prevent impairment of wound neoangiogenesis by diabetes. METHODS: Skin wounds were produced in the interscapular region of streptozotocin-induced diabetic mice. NGF (1 microg per day in PBS) or vehicle was applied onto the ulcers for 3 days after punching. Non-diabetic mice were used as controls. RESULTS: In wounds of untreated diabetic mice, endogenous levels of immunoreactive NGF were lower than those in wounds of non-diabetic mice ( p<0.01). Immunohistochemical analysis showed down-regulation of tyrosine kinase receptor-A (TrkA) and up-regulation of p75 receptor in granulation tissue microvasculature. Local NFG administration prevented diabetes-induced expressional alterations, enhanced reparative capillarisation ( p<0.01), and accelerated wound closure ( p<0.01). This was associated with a three-fold increase in endothelial cell proliferation ( p<0.01), while apoptosis was reduced by 50% ( p<0.05). Quantitative RT-PCR documented a 5.5-fold increase in the expression of vascular endothelial growth factor-A (VEGF-A) by exogenous NGF in diabetic tissues ( p<0.01). In in vitro preparations of human endothelial cells from derma, NGF increased the release of immunoreactive VEGF-A, and reduced high-glucose-induced apoptosis ( p<0.05), the latter effect being inhibited by a VEGF-A receptor-2 antagonist. CONCLUSIONS/INTERPRETATION: Diabetic ulcers display distinct alterations in reparative angiogenesis and in the expression of NGF and its receptors. NGF supplementation corrects endogenous liabilities, facilitates vascular regeneration, and suppresses endothelial apoptosis seemingly via VEGF-A. Our findings unravel new mechanisms responsible for NGF reparative action.

Persistent hepatic expression of human apo A-I after transfer with a helper-virus independent adenoviral vector.

Gene transfer with 'gutted' vectors is associated with persistent transgene expression and absence of hepatotoxicity, but the requirement of helper viruses hampers efficient production and leads to contamination of viral batches with these helper-viruses. In the present study, gene transfer with a helper-virus independent E(1)/E(3)/E(4)-deleted adenoviral vector induced persistent expression of human apo A-I (200 +/- 16 mg/dl at day 35, 190 +/- 15 mg/dl at 4 months, 170 +/- 16 mg/dl at 6 months) and stable transgene DNA levels (3.5 +/- 0.60 at day 35, 3.3 +/- 0.39 at 4 months, 3.1 +/- 0.47 mg/dl at 6 months) in C57BL/6 mice in the absence of significant toxicity. The vector contained the 1.5 kb human alpha(1)-antitrypsin promoter in front of the genomic human apo A-I sequence and four copies of the human apo E enhancer (hAAT.gA-I.4xapoE) and was deleted in E(1), E(3) and E(4). Reintroduction of E(4) ORF 3 and E(4) ORF 4 in the viral backbone caused a more than four-fold decline of transgene DNA between day 35 and 4 months after transfer both in wild-type and in C57BL/6 SCID and C57BL/6 Rag-1(-/-) mice, indicating that the effect of E(4) ORF 3 and E(4) ORF 4 is independent of a cellular immune response against viral epitopes. Co-injection of an E(1)-deleted vector containing no expression cassette and the E(1)/E(3)/E(4)-deleted vector containing the hAAT.gA-I.4xapoE expression cassette indicated that E(4) gene products destabilize transgene DNA in trans. Gene transfer with an E(1)/E(3)/E(4)-deleted vector containing only E(4) ORF 3 and the hAAT.gA-I.4xapoE expression cassette was associated with transgene DNA decline, but not with hepatotoxicity, indicating that transgene DNA persistence and hepatotoxicity are dissociated processes. After transfer with E(1)/E(3)/E(4)-deleted vectors containing expression cassettes with a different promoter or a different position of the apo E enhancers, transgene DNA levels were less stable than after transfer with the vector containing hAAT.gA-I.4xapoE, indicating that the expression cassette is an important determinant of episomal stability. In conclusion, gene transfer with an E(1)/E(3)/E(4)-deleted vector containing the hAAT.gA-I.4xapoE expression cassette induces persistent expression of human apo A-I in the absence of hepatotoxicity. Transgene DNA turnover is independent of an adaptive cellular immune response against viral epitopes and of hepatotoxicity. E(1)/E(3)/E(4)-deleted vectors containing transgenes under control of the hAAT promoter in combination with four copies of the human apo E enhancer may be suitable for hepatocyte-specific overexpression of transgenes after gene transfer. doi:10.1038/sj.gt.3301824

Sustained expression of human apo A-I following adenoviral gene transfer in mice.

Elevation of HDL cholesterol, following adenoviral apolipoprotein A-I (apo A-I) gene transfer, may delay or revert ischemic cardiovascular disease, provided transgene expression is persistent. The choice of promoter may have significant impact on persistence of transgene expression. Human apo A-I expression was compared after adenoviral gene transfer with a cytomegalovirus promoter (CMV) driven construct (AdCMV/A-I.gA-I) and with a construct (AdA-I.gA-I.4xapoE) containing the endogenous 256 bp apo A-I promoter (A-I), the genomic human apo A-I DNA (gA-I) and 4 human apo E enhancers (4xapoE) in three different mouse strains: C57BL/6, Balb/c and Fvb. After gene transfer with 5 x 10(8) p.f.u. of AdCMV/A-I.gA-I, human apo A-I expression was observed for 35 days in C57BL/6 mice, but declined below 1 mg/dl within 14 days both in Balb/c and Fvb mice, due to a strong humoral immune response against human apo A-I. In contrast, after transfer with AdA-I.gA-I.4xapoE, human apo A-I expression persisted for 6 months in all three strains and no antibodies against human apo A-I occurred in Fvb or Balb/c mice. Human apo A-I transgene DNA level 35 days after transfer with AdA-I.gA-I.4xapoE was 4.6- to 5.5-fold higher than with AdCMV/A-I.gA-I. CMV promoter attenuation occurred in all three strains, but promoter attenuation was not observed in any strain after transfer with AdA-I.gA-I.4xapoE. In conclusion, gene transfer with AdA-I.gA-I.4xapoE is associated with absence of an immune response against human apo A-I, improved transgene DNA persistence and absence of promoter shut-off, resulting in human apo A-I expression for up to 6 months in three different mouse strains. Possibly, the absence of human apo A-I expression in antigen-presenting cells with the liver-specific apo A-I promoter containing construct abrogated the immune response against human apo A-I in Balb/c and Fvb mice.

The Arg123-Tyr166 central domain of human ApoAI is critical for lecithin:cholesterol acyltransferase-induced hyperalphalipoproteinemia and HDL remodeling in transgenic mice.

High density lipoprotein (HDL) metabolism and lecithin:cholesterol acyltransferase (LCAT)-induced HDL remodeling were investigated in transgenic mice expressing human apolipoprotein (apo) AI or an apoAI/apoAII chimera in which the Arg123-Tyr166 domain of apoAI was substituted with the Ser12-Ala75 domain of apoAII. Expression of apoAI and of the apoAI/apoAII chimera resulted in a respective 3. 5-fold and 2.9-fold increase of HDL cholesterol. Human LCAT gene transfer into apoAI-transgenic mice resulted in a 5.1-fold increase of endogenous LCAT activity. This increase was associated with a 2. 4-fold increase of the cholesterol ester-to-free cholesterol ratio of HDL, a shift from HDL(3) to HDL(2), and a 2.4-fold increase of HDL cholesterol levels. Agarose gel electrophoresis revealed that human LCAT gene transfer into human apoAI-transgenic mice resulted in an increase of pre-beta-HDL and of pre-alpha-HDL. In contrast, human LCAT gene transfer did not affect cholesterol levels and HDL distribution profile in mice expressing the apoAI/apoAII chimera. Mouse LCAT did not "see" a difference between wild-type and mutant human apoAI, whereas human LCAT did, thus localizing the species-specific interaction in the central domain of apoAI. In conclusion, the Arg123-Tyr166 central domain of apoAI is not critical for in vivo lipoprotein association. It is, however, critical for LCAT-induced hyperalphalipoproteinemia and HDL remodeling independent of the lipid-binding properties of apoAI.

Sustained expression of human apolipoprotein A-I after adenoviral gene transfer in C57BL/6 mice: role of apolipoprotein A-I promoter, apolipoprotein A-I introns, and human apolipoprotein E enhancer.

Elevation of HDL cholesterol, after adenoviral apolipoprotein A-I (apo A-I) gene transfer, may delay or revert ischemic cardiovascular disease, provided transgene expression is persistent. Previously, we observed transient human apo A-I expression after adenoviral gene transfer with a cytomegalovirus (CMV)-driven construct containing the human apo A-I cDNA. Therefore, the effects of promoters (CMV or 256 base pairs of the human apo A-I promoter), introns of the human apo A-I gene, and the liver-specific human apolipoprotein E (apo E) enhancer on adenovirus-mediated human apo A-I expression were evaluated in C57BL/6 mice. In the presence of the CMV promoter, human apo A-I introns prolonged expression above 20 mg/dl from 14 to 35 days. Addition of one, two, or four copies of the human apo E enhancer in these constructs resulted in a copy-dependent but transient increase in expression for 14 days. The apo A-I promoter induced 3.2-fold lower peak levels of human apo A-I than did the CMV promoter, but insertion of four apo E enhancers in the apo A-I promoter-driven construct resulted in human apo A-I levels above 20 mg/dl for 6 months. The decline between day 6 and day 35 of human apo A-I expression driven by the CMV promoter was due to (1) a 2.5-fold decline in transgene DNA levels that is not observed with apo A-I promoter-driven constructs, and (2) CMV promoter attenuation as evidenced by a 7.6-fold decline in the human apo A-I mRNA/human apo A-I DNA copy number ratio between day 6 and day 35. Hepatotoxicity, as evidenced by up to 10-fold higher serum levels of transaminases on day 6 after gene transfer with CMV promoter-driven constructs than with apo A-I promoter-driven constructs, probably caused the accelerated decline of transgene DNA. In conclusion, gene transfer with an adenovirus comprising the 256-bp apo A-I promoter, the genomic apo A-I DNA, and four apo E enhancers, all of human origin, is associated with low hepatotoxicity and with the absence of promoter shutoff resulting in human apo A-I expression above 20 mg/dl for up to 6 months.