PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Curr Alzheimer Res. Author manuscript; available in PMC 2009 October 19.
Published in final edited form as:
PMCID: PMC2763541
NIHMSID: NIHMS131160

The Role of P-glycoprotein in Cerebral Amyloid Angiopathy; Implications for the Early Pathogenesis of Alzheimer’s Disease

Abstract

It has been shown in vitro that β-amyloid (Aβ) is transported by P-glycoprotein (P-gp). Previously, we demonstrated that Aβ immunoreactivity is significantly elevated in brain tissue of individuals with low expression of P-gp in vascular endothelial cells. These findings led us to hypothesize that P-gp might be involved in the clearance of Aβ in normal aging and particularly in Alzheimer’s disease (AD). As we were interested in the early pathogenesis of Aβ deposition, we studied the correlation between cerebral amyloid angiopathy (CAA) and P-gp expression in brain tissue samples from 243 non-demented elderly cases (aged 50 to 91 years). We found that endothelial P-gp and vascular Aβ were never colocalized, i.e., vessels with high P-gp expression showed no Aβ deposition in their walls, and vice versa. Aβ deposition occurred first in arterioles where P-gp expression was primarily low, and disappeared completely with the accumulation of Aβ. At this early stage, P-gp was upregulated in capillaries, suggesting a compensatory mechanism to increase Aβ clearance from the brain. Capillaries were usually affected only at later stages of CAA, at which point P-gp was lost even in these vessels. We hypothesize that Aβ clearance may be altered in individuals with diminished P-gp expression due, e.g., to genetic or environmental effects (such as drug administration). The impairment of Aβ clearance could lead to the accumulation and earlier deposition of Aβ, both in the walls of blood vessels and in the brain parenchyma, thus elevating the risk of CAA and AD.

Keywords: Alzheimer, apolipoprotein E, P-glycoprotein, cerebral amyloid angiopathy, risk factors, senile plaques, vascular amyloid, MDR1, degeneration

INTRODUCTION

Deposition of the β-amyloid peptide (Aβ) in the brain occurs in normal aging and is augmented in Alzheimer’s disease (AD) [1]. Aβ is found in senile plaques as well as in the walls of intracerebral and leptomeningeal blood vessels. Cerebral β-amyloid angiopathy (CAA) is one of the characteristic pathological features of AD, although the degree of CAA varies among cases [2,3]. In CAA, Aβ accumulates in the walls of arteries, arterioles and, to a lesser extent, of capillaries and veins [4]. Since cells of the brain continuously produce Aβ, it has been suggested that decreased clearance of the peptide from the brain at the blood brain barrier (BBB) could contribute to the buildup of Aβ [1,5,6], thereby increasing the likelihood of developing CAA and AD.

Recently, Lam et al. (2001) showed in vitro that Aβ is actively transported by P-glycoprotein (P-gp), a 170 kDa transmembrane protein that belongs to the ATP binding cassette (ABC) superfamily of transporters [7]. P-gp is the gene product of the multidrug resistance gene 1 (MDR1), which was initially discovered in tumor cells that had developed multidrug resistance during chemotherapy. It has since become evident that P-gp has important physiological functions. For example P-gp protects against xenobiotics, acting as a transporter in tissues with excretory and/or barrier function, such as intestine, liver, kidney and blood-brain-barrier [811].

P-gp is a major contributor to interindividual differences in drug response. Intestinal P-gp is inducible by St. John’s wort and rifampin, resulting in lower bioavailability of drugs such as digoxin or talinolol [1214]. In contrast, some drugs inhibit P-gp, and hence increase intestinal drug absorption; two examples are quinidine and verapamil [15,16]. However, little is known about the physiological function of P-gp in the central nervous system.

Considering the in vitro interaction of P-gp and Aβ [7], we analyzed brain samples from non-demented elderly humans and found an inverse correlation between the expression of vascular P-gp and the quantity of Aβ-positive senile plaques in the brain parenchyma, i.e., cases with low P-gp expression had high numbers of plaques, and vice versa [6]. In the present study, we extended our analysis to include the relationship between P-gp expression and the degree of β-amyloid angiopathy in this cohort.

MATERIALS AND METHODS

Tissue samples were obtained at autopsy from 243 subjects from northeastern Germany who died between the ages of 50 and 91 years [6,17,18] (Table 1). Because our goal was to elucidate the early events in the pathogenesis of AD-like pathology, we excluded late-stage cases with overt dementia from the analysis. Thus, we also were able to avoid ceiling effects associated with the pathology of late-stage AD. Aβ deposition and P-gp immunoreactivity were analyzed in the medial temporal lobe, including the hippocampus, dentate gyrus, subicular complex, Brodmann areas 28, 27, and 36, and superficial leptomeninges. The study was approved by the ethics committee of the medical faculty of the University of Greifswald.

Table 1
Sex and Age-range of the Analysed Cases

Immunohistochemistry

Immunohistochemical Aβ and P-gp quantification techniques have been described in detail elsewhere [6,17,18]. Briefly, Aβ was immunostained using polyclonal antibodies R163 and R165 to the C-terminal eight amino acids of Aβ40 and Aβ42, respectively (Pankaj Mehta, New York State Institute for Basic Research, Staten Island, New York). P-gp was detected with monoclonal antibody JSB-1 (Alexis, Gruenberg, Germany) using the automated system NexEs (Ventana Medical Systems, Inc. Frankfurt, Germany).

Leptomeningeal CAA was assessed using a rating scale of 0–4 that focused on the extent of amyloid deposition in the vascular wall in afflicted vessels (0 = no angiopathy, 1 = up to 25%, 2 = up to 50%, 3 = up to 75%, 4 = more than 75% of the circumference of all vessels affected). Within the parenchyma of the cortex and hippocampus, Aβ-positive vessels were counted.

For morphometric analysis of vascular P-gp expression in parenchyma, we assessed immunostained vessels in 15 high-magnification fields (0.2 mm2 per field) in the cerebral cortex, white matter and the hippocampus. P-gp expression was classified according to the number of positive vessels per total sampled area as high (20 or more positive vessels -score 3), mild (10–19 positive vessels - score 2), low (1–9 positive vessels - score 1) or negative (score 0).

For morphometric analysis of vascular P-gp expression in the superficial leptomeninges in 213 out of 243 cases, we assessed immunostained vessels in 6 high-magnification fields (0.2 mm2 per field). Samples with more than 10 immunopositive vessels were classified as strongly positive and given a P-gp index score of 3, samples with 6–10 positive vessels were classified as moderately positive (score 2), samples with 1–5 positive vessels were scored as low (score 1), and samples without staining were classified as negative (score 0).

In selected cases, confocal laser-scanning microscopy was performed using immunofluorescence double labeling with JSB-1 (dilution 1:100) for P-gp and polyclonal antibody Aβ42 (dilution 1:50, Chemicon, Germany) for Aβ. Slides were labeled with goat anti-mouse IgG (Alexa Fluor®, MoBiTec, Germany) for P-gp and goat anti-rabbit IgG (Alexa Fluor®; MoBiTec, Germany) for Aβ.

Genotyping

Apolipoprotein E genotype (243 cases) and polymorphisms in the MDR1 gene (240 out of 243 cases) were determined in DNA from fresh-frozen cerebellar tissue samples by PCR, as described before [6,1719].

Statistics

For statistical analysis, the Kruskal-Wallis test was used; P-values of odds ratios were calculated by the Chi2 test, and adjusted odds ratios were computed by logistic regression analysis, with age as confounding factor, using SPSS 9.0.

Results

CAA was found in 58 (Aβ40) or 55 (Aβ42), respectively, of the 243 cases analyzed. CAA occurred later than did Aβ plaques in the brain parenchyma; on average, subjects with CAA were 3.4–4 years older than those with Aβ plaques alone, and 7.3–8 years older than subjects without Aβ deposits (Table 2). If the cases with CAA were excluded, there was significantly lower expression of P-gp with age, i.e., cases younger than 70 years had significantly higher P-gp expression in hippocampus and cortex than did cases older than 70 years (P<0.05). Within the leptomeninges, a statistically significant decrement in P-gp expression was evident only in cases older than 80 years (P<0.05) (Fig. 1).

Fig. (1)
P-gp expression decreases significantly with age (cases without CAA). * P<0.05 (+/− standard error of mean)
Table 2
Mean Ages of Cohorts with Plaques and CAA Compared to Cases with Plaques Only. Note that the Mean Age of Cases with CAA is Greater than that of Cases with Senile Plaques in the Absence of CAA

Overall, the strongest immunostaining for P-gp was found in the endothelial cells of capillaries, whereas in arterioles of the brain and particularly of the leptomeninges, P-gp expression was significantly weaker or even non-detectable. Aβ deposition occurred first in the smooth muscle cell layer (tunica media) of arterioles, concomitant with the disappearance of P-gp immunoreactivity. At the same time, P-gp staining was significantly increased in capillaries of these cases (Fig. 2, ,3).3). In cases where CAA also involved capillaries, P-gp immunostaining was absent even in these small vessels. Consequently, in double-immunostained sections, Aβ and P-gp were never colocalized in the same vessel, i.e. vessels with Aβ deposits in their walls showed no P-gp, and vice versa (Fig. 4). There was no apparent difference between men and women in the expression of P-gp.

Fig. (2)
P-gp expression and Aβ40-CAA frequency in hippocampus, cortex and leptomeninges. * P<0.05
Fig. (3)
P-gp expression and Aβ42-CAA frequency in hippocampus, cortex and leptomeninges. * P<0.05
Fig. (4)
Fig. (4a and b). Double immunofluorescence of P-gp (green) and amyloid (red) with laser scanning microscopy. P-gp and CAA are not colocalized.

As described in our previous studies, homozygotes for apoEε4 had the lowest levels of P-gp expression and the highest vascular amyloid load [6,18]. The MDR1 polymorphisms in exon 2 G1A, exon 21 G2677T/A, as well as the putative functional exon 26 C3435T had no influence on CAA frequency (data not shown).

DISCUSSION

Cerebral β-amyloid angiopathy develops with increasing age, and can occur as a sporadic disorder [3,20], in inherited cases such as hereditary cerebral hemorrhage with amyloidosis of the Dutch type [21] and as a key feature of Alzheimer’s disease [3,22,23]. The impact of CAA on the development of AD and dementia has been underestimated in the past [24]. Despite the importance of CAA in AD and as a risk factor for stroke, the cellular and molecular events that lead to the deposition of Aβ in the vascular wall remain to be elucidated. Recent experimental data suggest that decreased clearance of soluble Aβ monomers at the BBB may contribute to the accumulation of Aβ in brain, and that P-gp acts as an efflux pump of Aβ at the blood brain barrier. Cultured cells that overexpress P-gp show increased elimination of Aβ [7]. In support of this finding, we found a significant inverse correlation between the number of Aβ-immunoreactive parenchymal plaques and the expression of endothelial P-gp in elderly humans [6]. Furthermore, there were significant differences in brain and plasma levels of Aβ40 and Aβ42 between MDR1a knock out mice and age-matched, wild-type controls [25].

In the present study, we observed differential expression of P-gp at different stages of Aβ deposition in cerebral blood vessels. In general, P-gp expression decreases with age, in parallel with increased numbers of plaques in the brain parenchyma. The effect of age on the function of the BBB is supported by a SPECT study in squirrel monkeys showing an age-related decline in the capacity of the BBB to remove Aβ from the brain that correlated with an increase in cerebrovascular Aβ40/42 load. The authors suggest that the pathogenic mechanism in CAA could involve reduced vascular clearance of Aβ [26]. However, in our study, the age-related decrease of P-gp expression was observed only in cases without CAA. As P-gp expression is upregulated in many capillaries when Aβ deposition occurs in other vessels, this reversal effect leads to a loss of statistical significance when all cases are analysed. The dynamic correlation between P-gp expression and CAA supports the view that P-gp plays an active role in the development of cerebral amyloidosis, and that changes in P-gp in the amyloidotic brain are not due solely to the aging process per se.

It has been generally accepted that CAA begins in larger, brain-supplying arteries, and then progresses into smaller vessels [24]. Our findings support this view. We also found that normal small arteries and arterioles of the cortex and leptomeninges revealed low or even undetectable P-gp, whereas capillaries usually show much stronger P-gp immunoreactivity, and remain free of Aβ in the first stages of CAA. In fact, we observed, for the first time, that in cases with early deposition of Aβ in arterioles, P-gp expression is significantly increased in capillaries. We hypothesize that the typically low expression of P-gp in arteries and arterioles contributes to the initial deposition of Aβ there; as β-amyloid is deposited in these vessels, P-gp increases compensatorily in capillaries in an attempt to reduce Aβ accumulation. As Aβ increases further, the capillaries themselves become amyloidotic and lose their P-gp, further exacerbating the buildup of Aβ in brain. This scheme predicts that Aβ in the blood should decrease as the as Aβ load in brain increases, owing to the deterioration of P-gp function in capillaries; indeed, in Tg2576 βAPP-transgenic mice, plasma Aβ levels decline as brain amyloid load grows [27].

The details of the interaction of Aβ and P-gp deserve further elucidation. So far it remains unclear why P-gp is completely lost in vessels with Aβ deposits. Endothelial cells appear to be preserved morphologically even after the smooth muscle cells have been disrupted by the accumulation of Aβ [4]. However, the toxic effects of Aβ lead to a disturbance of the function of the BBB [28,29], which may represent an additional effect of Aβ on local P-gp expression. Whether or not the correlative changes in CAA and P-gp represent a causal link remains to be determined. The fact that P-gp is never found in vessels with Aβ strongly favors the idea of a pathogenic interrelationship, either alone or (more likely) involving other transporters as well.

As apoEε4 genotype correlates with increased deposition of Aβ, we found also a correlation between apoE-genotype and P-gp expression [6]. Specifically, elderly homozygotes for apoEε4 had the greatest Aβ load and the lowest levels of vascular P-gp expression. The nature of the interaction between apoE type and P-gp expression also requires further exploration.

The P-gp story is intriguing for two reasons in particular: 1) P-gp can be pharmacologically up- and down-regulated. The importance of this problem became recognized in connection with chemotherapy for malignant tumors, in particular for malignant brain tumors, where up-regulation of P-gp leads to drug resistance [30]. Intestinal P-gp is inducible by rifampin, St. John’s wort and thyroxine, resulting in lower bioavailability of drugs such as digoxin or talinolol [1214]. In contrast, some drugs such as quinidine [15], verapamil [16] and talinolol [31] inhibit P-gp and hence increase intestinal drug absorption; 2) P-gp expression may vary in different individuals due to genetic polymorphisms [19, 32]. Recently, it was shown that the human MDR1 (ABCB1) gene is highly polymorphic [19, 33]. To date, more than 16 single nucleotide polymorphisms (SNPs) have been identified. However, an alteration of transport function could be demonstrated only for the silent mutation in exon 26. This C3435T SNP correlates with the level of P-gp expression in the intestine, where the C-allele was found to be associated with increased P-gp levels [19, 32].

In summary, our results support the hypothesis that P-gp plays a role in the clearance of Aβ at the BBB. Thus, low P-gp expression in the brain vasculature might represent a new risk factor for the cerebral proteopathies. Furthermore, selective augmentation of P-gp function in brain could be a novel strategy for the prevention of Aβ accumulation in brain, thereby diminishing the likelihood of developing CAA and AD.

Acknowledgments

We thank D. Wegner for his support regarding the statistical analysis, as well as C. Müller, S. Uffmann, A. Wolter, and I. Geissler for excellent technical assistance.

Biography

An external file that holds a picture, illustration, etc.
Object name is nihms131160b1.gif

References

1. Selkoe DJ. Translating cell biology into therapeutic advances in Alzheimer’s disease. Nature. 1999;399:A23–A31. [PubMed]
2. Olichney JM, Hansen LA, Hofstetter CR, Grundman M, Katzman R, Thal LJ. Cerebral infarction in Alzheimer’s disease is associated with severe amyloid angiopathy and hypertension. Arch Neurol. 1995;52:702–8. [PubMed]
3. Jellinger KA. Alzheimer disease and cerebrovascular pathology: an update. J Neural Transm. 2002;109:813–36. [PubMed]
4. Revesz T, Holton JL, Lashley T, Plant G, Rostagno A, Ghiso J, Frangione B. Sporadic and familial cerebral amyloid angiopathies. Brain Pathol. 2002;12:343–357. [PubMed]
5. Shibata M, Yamada S, Kumar SR, Calero M, Bading J, Frangione B, Holtzman DM, Miller CA, Strickland DK, Ghiso J, Zlokovic BV. Clearance of Alzheimer’s amyloid-β 1–40 peptide from brain by LDL receptor-related protein-1 at the blood-brain barrier. J Clin Invest. 2000;106:1489–1499. [PMC free article] [PubMed]
6. Vogelgesang S, Cascorbi I, Schroeder E, Pahnke J, Kroemer HK, Siegmund W, Kunert-Keil C, Walker LC, Warzok RW. Deposition of Alzheimer’s beta-amyloid is inversely correlated with P-glycoprotein expression in the brains of elderly non-demented humans. Pharmacogenetics. 2002;12:535–41. [PubMed]
7. Lam FC, Liu R, Shapiro AB, Renoir JM, Sharom FJ, Reiner PB. β-amyloid efflux mediated by p-glycoprotein. J Neurochem. 2001;76:1121–1128. [PubMed]
8. Fromm MF, Kaufmann HM, Fritz P, Burk O, Kroemer HK, Warzok RW, Eichelbaum M, Siegmund W, Schrenk D. The effect of rifampin treatment on intestinal expression of human MRP transporters. Am J Pathol. 2000;157:1575–1580. [PubMed]
9. Fromm MF. P-glycoprotein: a defense mechanism limiting oral bioavailability and CNS accumulation of drugs. Int J Clin Pharmacol Ther. 2000;38:69–74. [PubMed]
10. Schinkel AH. The physiological function of drug transporting P-glycoproteins. Semin Cancer Biol. 1997;8:161–170. [PubMed]
11. Zhang Y, Benet LZ. The gut as a barrier to drug absorption. Combined role of cytochrome P4503A and P-glycoprotein. Clin Pharmacokinet. 2001;40:159–168. [PubMed]
12. Greiner B, Eichelbaum M, Fritz P, Kreichgauer HP, von Richter O, Zundler J, Kroemer HK. The role of intestinal P-glycoprotein in the interaction of digoxin and rifampin. J Clin Invest. 1999;104:147–153. [PMC free article] [PubMed]
13. Westphal K, Weinbrenner A, Zschiesche M, Franke G, Knoke M, Oertel R, Fritz P, Richter O, Warzok R, Hachenberg T, Kauffmann HM, Schrenk D, Terhaag B, Kroemer HK, Siegmund W. Induction of p-glycoprotein by rifampin increases intestinal secretion of talinolol in human beings: a new type of drug/drug interaction. Clin Pharmacol Ther. 2000;68:345–355. [PubMed]
14. Johne A, Brockmoeller J, Bauer S, Maurer A, Langheinrich M, Roots I. Pharmacokinetic interaction of digoxin with an herbal extract from St John’s wort (Hypericum perforatum) Clin Pharmacol Ther. 1999;66:338–345. [PubMed]
15. Fromm MF, Kim RB, Stein CM, Wilkinson GR, Roden DM. Inhibition of P-glycoprotein-mediated drug transport: A unifying mechanism to explain the interaction between digoxin and quinidine. Circulation. 1999;99:552–557. [PubMed]
16. Gramatté T, Oertel R. Intestinal secretion of intravenous talinolol is inhibited by luminal R-verapamil. Clin Pharmacol Ther. 1999;66:239–245. [PubMed]
17. Warzok R, Kessler C, Apel G, Schwarz A, Egensperger R, Schreiber D, Herbst EW, Wolf E, Walther R, Walker LC. Apolipoprotein E promotes incipient Alzheimer pathology in the elderly. Alzheimer Dis Rel Dis. 1998;12:33–39. [PubMed]
18. Walker LC, Pahnke J, Madauß M, Vogelgesang S, Pahnke A, Herbst EW, Stausske D, Walther R, Kessler C, Warzok RW. Apolipoprotein E4 promotes the early deposition of Aβ42 and then Aβ 40 in the elderly. Acta Neuropathol. 2000;100:36–42. [PubMed]
19. Hoffmeyer S, Burk O, von Richter O, Arnold HP, Brockmöller Johne A, Cascorbi I, Gerloff T, Roots I, Eichelbaum M, Brinkmann U. Functional polymorphisms of the human multidrug resistance gene: multiple sequence variations and correlation with P-glycoprotein expression and activity in vivo. Proc Natl Acad Sci. 2000;97:3473–3478. [PubMed]
20. Esiri MM, Wilcock GK. Cerebral amyloid angiopathy in dementia and old age. J Neurol Neurosurg Psychiatry. 1986;49:1221–1226. [PMC free article] [PubMed]
21. Levy E, Carman MD, Fernandez-Madrid IJ, Power MD, Lieberburg I, van Duinen SG, Bots GT, Luyendijk W, Frangione B. Mutation of the Alzheimer’s disease amyloid gene in hereditary cerebral hemorrhage, Dutch type. Science. 1990;248:1124–6. [PubMed]
22. Bergeron C, Ranalli PJ, Miceli PN. Amyloid angiopathy in Alzheimer’s disease. Can J Neurol Sci. 1987;14:564–9. [PubMed]
23. Vinters HV, Vonsattel JP. Neuropathologic features and grading of Alzheimer-related and sporadic CAA. In: Verbeek MM, de Waal RMW, Vinters HV, editors. Cerebral Amyloid Angiopathy in Alzheimer’s Disease and Related Disorders. Kluwer, Dordrecht: 2000. pp. 137–155.
24. Rensink AAM, de Waal RMW, Kremer B, Verbeek MM. Pathogenesis of cerebral amyloid angiopathy. Brain Res Rev. 2003;43:207–23. [PubMed]
25. Wu S, Webb D, Dodart JC, DeMattos RB, Paul SM, Bales KR, Holtzman DM. Reduced brain and elevated plasma A40 levels in the P-glycoprotein (MDR1A) knockout mouse. Washington, DC: Society for Neuroscience; 2003. Program No. 731.3. 2003 Abstract Viewer/Itinerary Planner. Online.
26. Bading JR, Yamada S, Mackic JB, Kirkman L, Miller C, Calero M, Ghiso J, Frangione B, Zlokovic BV. Brain clearance of Alzheimer’s amyloid-beta40 in the squirrel monkey: a SPECT study in a primate model of cerebral amyloid angiopathy. J Drug Target. 2002;10:359–68. [PubMed]
27. Kawarabayashi T, Younkin LH, Saido TC, Shoji M, Ashe KH, Younkin SG. Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer’s disease. J Neurosci. 2001;21:372–81. [PubMed]
28. Frangione B, Revesz T, Vidal R, Holton J, Lashley T, Houlden H, Wood N, Rostagno A, Plant G, Ghiso J. Familial cerebral amyloid angiopathy related to stroke and dementia. Amyloid. 2001;8 (Suppl 1):36–42. [PubMed]
29. Miravalle L, Tokuda T, Chiarle R, Giaccone G, Bugiani O, Tagliavini F, Frangione B, Ghiso J. Substitutions at codon 22 of Alzheimer’s abeta peptide induce diverse conformational changes and apoptotic effects in human cerebral endothelial cells. J Biol Chem. 2000;275:27110–6. [PubMed]
30. Baehr O, Rieger J, Duffner F, Meyermann R, Weller M, Wick W. P-glycoprotein and multidrug resistance-associated protein mediate specific patterns of multidrug resistance in malignant glioma cell lines, but not in primary glioma cells. Brain Pathol. 2003;13:482–494. [PubMed]
31. Westphal K, Weinbrenner A, Gießmann T, Stuhr M, Franke G, Zschiesche M, Oertel R, Terhaag B, Kroemer HK, Siegmund W. Oral bioavailability of digoxin is enhanced by talinolol: Evidence for involvement of intestinal P-glycoprotein. Clin Pharmacol Ther. 2000;68:6–12. [PubMed]
32. Cascorbi I, Gerloff T, Johne A, Meisel C, Hoffmeyer S, Schwab M, Schaeffeler E, Eichelbaum M, Brinkmann U, Roots I. Frequency of single nucleotide polymorphisms in the P-glycoprotein drug transporter MDR1 gene in white subjects. Clin Pharmacol Ther. 2001;69:169–174. [PubMed]
33. Brinkmann U, Roots I, Eichelbaum M. Pharmacogenetics of the human drug- transporter gene MDR1: impact of polymorphisms on pharmacotherapy. DDT. 2001;6:835–839. [PubMed]