Journal of Medical Virology 30:249-252 (1990)
Case Report: Electron Microscopic Detection of Human Parvovirus (B19) in a Patient With HIV Infection Ian L. Chrystie, June D. Almeida, and Jan Welch Department of Virology, United Medical Schools of Guy’s and S t . Thomas’s, S t . Thomas’s Campus (I.L.C., J.D.A.), and Department of Genitourinary Medicine, S t . Thomas’s Hospital Id.W.), London, England
We report the findings on an HIV-positive patient found to be infected with human parvovirus B19. In a comparative study of sequential sera, direct electron microscopy (EM) had the same sensitivity as DNA hydridisation for the detection of the virus. EM did not require specific reagents and also yielded information on the state of the virus; i.e., whether random or complexed, and the type of antibody involved. The presence of parvovirus complicated the judgement as to drug treatment. This case also highlighted the importance of diagnosing the presence of opportunistic viral infections that have no, or lowgrade, pathogenic activity.
KEY WORDS: AIDS, electron microscopy, diagnostic virology
INTRODUCTION Human parvovirus B19 was first discovered in the serum of apparently healthy blood donors in 1975 by electrophoresis and electromicroscopy (EM) I Cossart et al., 19751. It was later demonstrated to be associated with such conditions as aplastic crises in patients with chronic haemolytic anaemia [e.g., Pattison et al., 1981I, erythema infectiosum I Anderson et al., 1983 I, hydrops foetalis [Brown et al., 19841, rheumatoid-like arthritis [White et al., 19851, and chronic suppression of the bone marrow of immunocompromised patients leading to red cell aplasia IKurtzman et al., 19871. At present, DNA hybridisation and detection of specific IgM are the most frequently employed methods for the detection of parvovirus infection [Anderson et al., 1985; Cohen et al., 19831. This paper reports a case of red cell aplasia associated with human parvovirus infection in a patient with AIDS, and demonstrates that EM and immune EM (IEM) still provide a sensitive and informative technique for the detection and identification of parvoviraemia. Since EM allows direct visualisation of the virus it may also provide information as to whether the virus is present in the serum as single, unaggre1990 WILEY-LISS, INC.
gated particles or in the form of immune complexes; even further, it can detect the presence of IgM and distinguish it from IgG in the complexes [Almeida and Waterson, 19691. In the case we present, examination of sequentially obtained sera revealed that both single and complexed particles were present in the serum and that clearly recognisable IgM could be visualised within the complexes. The case also illustrates that parvovirus infection can complicate the assessment of antiviral therapy.
MATERIALS AND METHODS Patient A 26 year-old heterosexual Ugandan male was found to be HIV positive in May 1988, and 2 months later he developed Pneumocystis carinii pneumonia. He defaulted from follow-up until September 1988, when he presented complaining of tiredness and was found to be severely anaemic (haemoglobin 3.1 gidl). Bone marrow examination revealed red cell aplasia. During the following 2 months the patient remained reticulocytopenic and required repeated blood transfusions. In December he was treated with intravenous immunoglobulin (Gammimune N; 300 mglkgiday for 8 days) following which he developed a reticulocytosis and his haemoglobin rose to normal levels. In February 1989 he was started on zidovudine; 6 weeks later his haemoglobin level fell (to 7.0 gidl) and he became reticulocytopenic. Although zidovudine was stopped the patient remained anaemic and by May 1989 was again transfusion dependent. Following similar treatment with intravenous immunoglobulin in June 1989 a reticulocytosis and a rise in haemoglobin to 9.0 gldl occurred [Mitchell et al., 19901. From September 1988 to June 1989 serum samples were collected and levels of parvovirus DNA and par-
Accepted for publication December 22, 1989. Address reprint requests to Dr. Ian L. Chrystie, Department of Virology, United Medical Schools of Guy’s and St. Thomas’s, St. Thomas Campus, London SE1 7EH, England.
Chrystie et al.
250
TABLE I. Detection of Parvovirus by Electron Microscopy, DNA Hybridisation, and Serology on Sequential Sera From a n HIV-Positive Patient Sequential dates EM" 040778-011288 CLUMPSiSNGLS'iIgM 040788 CLUMPSiSNGLSiIeM 120788 140988 CLUMPS~SNGLS~I~M CLUMPSiSNGLSiIgM 021188 011288 SNGLSiRARE CLUMPS Immunoglobulin given for 10 days 131288-060689 131288 Negative 161288 Negative 211288 Negative 301288 Negative 010289 Negative 080389 Few SNGLS 150389 Negative 050489 CLUMPSiSNGLSiIgM 060689 CLUMPSiSNGLSiIgM Immunoglobulin given for 10 days 290689-190789 290689 CLUMPSiSNGLSiIgM 070789 Negative 120789 Negative 190789 Negative
DNA"
IgM'
IgG'
ND' ND
ND ND
ND ND
+ +
I
-
+-
~
~
~
~
ND ~
~
+ ~
I
+
~
~
ND ~
~
~
+-
ND
ND ND
ND ND
~
+ I
ND
+ + + + ND + + + ~
ND i
ND ND
"Negative staining. hDNA hybridisation. 'Radioimmunoassay. %ingle particles without Ab. ''Test not requested.
vovirus-specific IgG and IgM were determined a t the Department of Microbiology, University College Hospital, London [Anderson et al., 1982,19851. These samples were later examined by EM as described below, and the results are presented in Table I.
Electron Microscopy Aliquots (0.25 ml) of serum were diluted to 5 ml with phosphate-buffered saline (PBS) and centrifuged a t 100,000G for 1hr. The resultant pellet was resuspended in 0.25 ml of PBS and examined by negative staining EM by using 3% sodium phosphotungstate a t pH 6.5. For immune EM 0.05 ml of the serum pellet was incubated for 1 h r at room temperature with a n equal amount of a human parvovirus antiserum (kindly provided by Dr. B. Cohen, PHLS, Colindale, London). A drop of this suspension was transferred to a grid, washed with several drops of sterile distilled water, and stained with phosphotungstate as before.
RESULTS The results of analysis of the sera for parvovirus by DNA hybridisation and parvovirus-specific immunoglobulins are shown in Table I. Specific IgG was present from 9188 to 4189 and was detected a t a very low level in 7/89. Virus-specific IgM was only detected, at low concentration, in 11/88.Virus DNA was detected in two periods from 9 to 12188 and 4 to 6189.
Parvovirus was detected by EM in serum samples from 7/88 to early 12/88. The earlier samples contained virus for the most part in aggregated form (Fig. l),with only occasional single particles (Fig. 2). In the later samples this pattern was reversed with the virus appearing mainly unaggregated although some small clumps were still visible. Virus was again detected in samples taken between 4189 and 6189. In these samples the vast majority of the virus was in unaggregated form. An additional feature of the virudantibody aggregates was that in several instances it was possible to visualise IgM molecules (see Table I). A complex illustrating this feature is shown in Figure 1. Detection of virus-specific DNA paralleled visualisation of virus by EM. The results of IEM confirmed that the virus was clumped by the anti-B19 serum used. This was, of course, easiest to appreciate when the the method was applied to those sera which contained only single particles. However, in no instance was a sample IEM positive and EM negative.
DISCUSSION Although the results we present are from only a single HIV-positive patient they are of interest from several different viewpoints. Firstly, and most simply, they illustrate that direct, negative staining of serum pellets proved to be a s sen-
EM Detection of Human Parvovirus B19
251
Fig. 1.Typical micrograph of parvovirus present in the serum of the patient under investigation. The virus is clearly aggregated and it is possible to resolve antibody molecules linking the particles. These are mainly IgG but distinct IgM molecules are also present (see arrowheads). Bar = 50 nm.
Fig. 2. Throughout the sequential EM study the distribution of particles changed between mainly aggregated and mainly single. This micrograph shows the stage of numerous small aggregates alongside single particles. Bar- 100 nm.
sitive as DNA hybridisation for the detection of parvovirus infection in this patient. This in spite of the fact that DNA hybridisation is estimated to detect lo4 particles per ml IAnderson et al., 19851 whereas EM is generally accepted to be limited to lo6 particles per ml. Since the EM-negative staining technique is a general method of rapid virus diagnosis, it may be used a s a means of parvovirus detection without needing t o employ specific reagents such as antisera or radioactive nucleic acid probes. Thus EM may circumvent the need for a specific parvovirus test. In this context, although specific antiserum used in IEM established the identity of the particles it did not increase the sensitivity of the test. Secondly, EM studies enable investigators t o determine whether the virus is present in the unaggregated or complexed form. Micrographs taken a t the time of initial detection of the virus showed the virus to be mostly in the form of immune complexes, but, as the patient deteriorated, more unaggregated particles were seen, although some clumps were still present. At this point pooled human gamma globulin was given and the virus could not be detected for the subsequent period. These findings are not easy to interpret. Since the patient had AIDS the progression from complexed to unaggregated virus could reflect a change in the
level of specific immunoglobulin. Alternatively i t could be related to the pathogenesis of parvovirus infection with the virus doing most damage when circulating complexes were present. Further study of additional cases would be needed to clarify these points. IgM was detected more frequently by EM than by the specific IgM RIA. This could possibly be explained by the fact that the IgM is complexed and not available in free form. Perhaps the most important aspect of this case is that the parvovirus infection may well have complicated the assessment of treatment with zidovudine. Drug treatment was stopped when anaemia recurred although this did not restore his haemoglobin to normal levels. However, during each anaemic crisis haemopoiesis was restored, and observable parvoviraemia was cleared by the use of intravenous, pooled gamma globulin which invariably contains antibody to the virus. Parvovirus infection is not uncommon, 61%of London blood donors having antibody to the virus LCohen et al., 19831, and parvovirus should therefore be considered as a possible cause of red cell aplasia in immunocompromised patients. Finally, from a n epidemiological viewpoint, the presence of increasing numbers of immunocompromised individuals within the community, both HIV related and
Chrystie et al.
252
as a result of immune manipulation of a large number of diseases, is providing a n enlarged host reservoir for opportunistic microbial agents. These are readily detected if they cause disease, and there is now considerable literature on the opportunistic infections, both bacterial and viral, that afflict the immunocompromised. However, there could also be a n increasing number of infections by organisms whose effect is less obvious: this group could well contain latent or orphan viruses in addition to such viruses as parvovirus and the polyoma virus family where the relationship of aetiological cause and disease effect is less well understood.
ACKNOWLEDGMENTS We wish to thank Professor J.R. Pattison and his colleagues for their helpful collaboration in carrying out the DNA hybridisation and serological tests. REFERENCES Almeida JD, Waterson AP (1969): The morphology of virus-antibody interaction. Advances in Virus Research 15:307-338. Anderson MJ, Davis LR, Jones SE, Pattison JR, Serjeant GR (1982): The development and use of a n antibody capture radioimmuno-
assay for specific IgM to a human parvovirus-like agent. Journal of Hygiene (Cambridge) 88:309-324. Anderson MJ, Jones SE, Fisher-Hoch SP, Lewis E, Hall SM, Bartlett CLR, Cohen BJ, Mortimer PP, Pereira MS (1983): Human parvovirus, the cause of erythema infectiosum (fifth disease). Lancet 1:1378. Anderson MJ, Jones SE, Minson AC (1985): Diagnosis of human parvovirus infection by dot-blot hybridisation using cloned viral DNA. Journal of Medical Virology 15:163-172. Brown T, Anand A, Ritchie LD, Clewley J P , Reid TM (1984): Intrauterine parvovirus infection associated with hydrops fetalis. Lancet 2:1033-1034. Cohen BJ, Mortimer PP, Pereira MS (1983): Diagnostic assays with monoclonal antibodies for the human serum parvovirus-like virus (SPLV1. Journal of Hygiene (Cambridge1 9 1:113 -130. Cossart YE, Field AM, Cant B, Widdows D (1975): Parvovirus-like particles in human sera. Lancet 1:72-73. Kurtzman GJ, Ozowa K, Cohen B, Hanson GR, Oseas R, Young NS (1987): Chronic bone marrow failure due to persistent B19 parvovirus infection. New England Journal of Medicine 317:287294. Mitchell S, Welch J , Nicholson F, Weston-Smith S, Bradbeer CS (1990):Parvovirus infection and anaemia in a patient with AIDS: Case report. Genitourinary Medicine (in press). Pattison J R , Jones SE, Hodgson J, Davis LR, White JM, Stroud CE, Murtaza L (1981): Parvovirus infections and hypoplastic crisis in sickle-cell anaemia. Lancet 1:664-665. White DG, Woolf AD, Mortimer PP, Cohen BJ, Blake DR, Bacon PA (1985):Human parvovirus arthropathy. Lancet 1:419-421.
Case Report: Electron Microscopic Detection of Human Parvovirus (B19) in a Patient With HIV Infection Ian L. Chrystie, June D. Almeida, and Jan Welch Department of Virology, United Medical Schools of Guy’s and S t . Thomas’s, S t . Thomas’s Campus (I.L.C., J.D.A.), and Department of Genitourinary Medicine, S t . Thomas’s Hospital Id.W.), London, England
We report the findings on an HIV-positive patient found to be infected with human parvovirus B19. In a comparative study of sequential sera, direct electron microscopy (EM) had the same sensitivity as DNA hydridisation for the detection of the virus. EM did not require specific reagents and also yielded information on the state of the virus; i.e., whether random or complexed, and the type of antibody involved. The presence of parvovirus complicated the judgement as to drug treatment. This case also highlighted the importance of diagnosing the presence of opportunistic viral infections that have no, or lowgrade, pathogenic activity.
KEY WORDS: AIDS, electron microscopy, diagnostic virology
INTRODUCTION Human parvovirus B19 was first discovered in the serum of apparently healthy blood donors in 1975 by electrophoresis and electromicroscopy (EM) I Cossart et al., 19751. It was later demonstrated to be associated with such conditions as aplastic crises in patients with chronic haemolytic anaemia [e.g., Pattison et al., 1981I, erythema infectiosum I Anderson et al., 1983 I, hydrops foetalis [Brown et al., 19841, rheumatoid-like arthritis [White et al., 19851, and chronic suppression of the bone marrow of immunocompromised patients leading to red cell aplasia IKurtzman et al., 19871. At present, DNA hybridisation and detection of specific IgM are the most frequently employed methods for the detection of parvovirus infection [Anderson et al., 1985; Cohen et al., 19831. This paper reports a case of red cell aplasia associated with human parvovirus infection in a patient with AIDS, and demonstrates that EM and immune EM (IEM) still provide a sensitive and informative technique for the detection and identification of parvoviraemia. Since EM allows direct visualisation of the virus it may also provide information as to whether the virus is present in the serum as single, unaggre1990 WILEY-LISS, INC.
gated particles or in the form of immune complexes; even further, it can detect the presence of IgM and distinguish it from IgG in the complexes [Almeida and Waterson, 19691. In the case we present, examination of sequentially obtained sera revealed that both single and complexed particles were present in the serum and that clearly recognisable IgM could be visualised within the complexes. The case also illustrates that parvovirus infection can complicate the assessment of antiviral therapy.
MATERIALS AND METHODS Patient A 26 year-old heterosexual Ugandan male was found to be HIV positive in May 1988, and 2 months later he developed Pneumocystis carinii pneumonia. He defaulted from follow-up until September 1988, when he presented complaining of tiredness and was found to be severely anaemic (haemoglobin 3.1 gidl). Bone marrow examination revealed red cell aplasia. During the following 2 months the patient remained reticulocytopenic and required repeated blood transfusions. In December he was treated with intravenous immunoglobulin (Gammimune N; 300 mglkgiday for 8 days) following which he developed a reticulocytosis and his haemoglobin rose to normal levels. In February 1989 he was started on zidovudine; 6 weeks later his haemoglobin level fell (to 7.0 gidl) and he became reticulocytopenic. Although zidovudine was stopped the patient remained anaemic and by May 1989 was again transfusion dependent. Following similar treatment with intravenous immunoglobulin in June 1989 a reticulocytosis and a rise in haemoglobin to 9.0 gldl occurred [Mitchell et al., 19901. From September 1988 to June 1989 serum samples were collected and levels of parvovirus DNA and par-
Accepted for publication December 22, 1989. Address reprint requests to Dr. Ian L. Chrystie, Department of Virology, United Medical Schools of Guy’s and St. Thomas’s, St. Thomas Campus, London SE1 7EH, England.
Chrystie et al.
250
TABLE I. Detection of Parvovirus by Electron Microscopy, DNA Hybridisation, and Serology on Sequential Sera From a n HIV-Positive Patient Sequential dates EM" 040778-011288 CLUMPSiSNGLS'iIgM 040788 CLUMPSiSNGLSiIeM 120788 140988 CLUMPS~SNGLS~I~M CLUMPSiSNGLSiIgM 021188 011288 SNGLSiRARE CLUMPS Immunoglobulin given for 10 days 131288-060689 131288 Negative 161288 Negative 211288 Negative 301288 Negative 010289 Negative 080389 Few SNGLS 150389 Negative 050489 CLUMPSiSNGLSiIgM 060689 CLUMPSiSNGLSiIgM Immunoglobulin given for 10 days 290689-190789 290689 CLUMPSiSNGLSiIgM 070789 Negative 120789 Negative 190789 Negative
DNA"
IgM'
IgG'
ND' ND
ND ND
ND ND
+ +
I
-
+-
~
~
~
~
ND ~
~
+ ~
I
+
~
~
ND ~
~
~
+-
ND
ND ND
ND ND
~
+ I
ND
+ + + + ND + + + ~
ND i
ND ND
"Negative staining. hDNA hybridisation. 'Radioimmunoassay. %ingle particles without Ab. ''Test not requested.
vovirus-specific IgG and IgM were determined a t the Department of Microbiology, University College Hospital, London [Anderson et al., 1982,19851. These samples were later examined by EM as described below, and the results are presented in Table I.
Electron Microscopy Aliquots (0.25 ml) of serum were diluted to 5 ml with phosphate-buffered saline (PBS) and centrifuged a t 100,000G for 1hr. The resultant pellet was resuspended in 0.25 ml of PBS and examined by negative staining EM by using 3% sodium phosphotungstate a t pH 6.5. For immune EM 0.05 ml of the serum pellet was incubated for 1 h r at room temperature with a n equal amount of a human parvovirus antiserum (kindly provided by Dr. B. Cohen, PHLS, Colindale, London). A drop of this suspension was transferred to a grid, washed with several drops of sterile distilled water, and stained with phosphotungstate as before.
RESULTS The results of analysis of the sera for parvovirus by DNA hybridisation and parvovirus-specific immunoglobulins are shown in Table I. Specific IgG was present from 9188 to 4189 and was detected a t a very low level in 7/89. Virus-specific IgM was only detected, at low concentration, in 11/88.Virus DNA was detected in two periods from 9 to 12188 and 4 to 6189.
Parvovirus was detected by EM in serum samples from 7/88 to early 12/88. The earlier samples contained virus for the most part in aggregated form (Fig. l),with only occasional single particles (Fig. 2). In the later samples this pattern was reversed with the virus appearing mainly unaggregated although some small clumps were still visible. Virus was again detected in samples taken between 4189 and 6189. In these samples the vast majority of the virus was in unaggregated form. An additional feature of the virudantibody aggregates was that in several instances it was possible to visualise IgM molecules (see Table I). A complex illustrating this feature is shown in Figure 1. Detection of virus-specific DNA paralleled visualisation of virus by EM. The results of IEM confirmed that the virus was clumped by the anti-B19 serum used. This was, of course, easiest to appreciate when the the method was applied to those sera which contained only single particles. However, in no instance was a sample IEM positive and EM negative.
DISCUSSION Although the results we present are from only a single HIV-positive patient they are of interest from several different viewpoints. Firstly, and most simply, they illustrate that direct, negative staining of serum pellets proved to be a s sen-
EM Detection of Human Parvovirus B19
251
Fig. 1.Typical micrograph of parvovirus present in the serum of the patient under investigation. The virus is clearly aggregated and it is possible to resolve antibody molecules linking the particles. These are mainly IgG but distinct IgM molecules are also present (see arrowheads). Bar = 50 nm.
Fig. 2. Throughout the sequential EM study the distribution of particles changed between mainly aggregated and mainly single. This micrograph shows the stage of numerous small aggregates alongside single particles. Bar- 100 nm.
sitive as DNA hybridisation for the detection of parvovirus infection in this patient. This in spite of the fact that DNA hybridisation is estimated to detect lo4 particles per ml IAnderson et al., 19851 whereas EM is generally accepted to be limited to lo6 particles per ml. Since the EM-negative staining technique is a general method of rapid virus diagnosis, it may be used a s a means of parvovirus detection without needing t o employ specific reagents such as antisera or radioactive nucleic acid probes. Thus EM may circumvent the need for a specific parvovirus test. In this context, although specific antiserum used in IEM established the identity of the particles it did not increase the sensitivity of the test. Secondly, EM studies enable investigators t o determine whether the virus is present in the unaggregated or complexed form. Micrographs taken a t the time of initial detection of the virus showed the virus to be mostly in the form of immune complexes, but, as the patient deteriorated, more unaggregated particles were seen, although some clumps were still present. At this point pooled human gamma globulin was given and the virus could not be detected for the subsequent period. These findings are not easy to interpret. Since the patient had AIDS the progression from complexed to unaggregated virus could reflect a change in the
level of specific immunoglobulin. Alternatively i t could be related to the pathogenesis of parvovirus infection with the virus doing most damage when circulating complexes were present. Further study of additional cases would be needed to clarify these points. IgM was detected more frequently by EM than by the specific IgM RIA. This could possibly be explained by the fact that the IgM is complexed and not available in free form. Perhaps the most important aspect of this case is that the parvovirus infection may well have complicated the assessment of treatment with zidovudine. Drug treatment was stopped when anaemia recurred although this did not restore his haemoglobin to normal levels. However, during each anaemic crisis haemopoiesis was restored, and observable parvoviraemia was cleared by the use of intravenous, pooled gamma globulin which invariably contains antibody to the virus. Parvovirus infection is not uncommon, 61%of London blood donors having antibody to the virus LCohen et al., 19831, and parvovirus should therefore be considered as a possible cause of red cell aplasia in immunocompromised patients. Finally, from a n epidemiological viewpoint, the presence of increasing numbers of immunocompromised individuals within the community, both HIV related and
Chrystie et al.
252
as a result of immune manipulation of a large number of diseases, is providing a n enlarged host reservoir for opportunistic microbial agents. These are readily detected if they cause disease, and there is now considerable literature on the opportunistic infections, both bacterial and viral, that afflict the immunocompromised. However, there could also be a n increasing number of infections by organisms whose effect is less obvious: this group could well contain latent or orphan viruses in addition to such viruses as parvovirus and the polyoma virus family where the relationship of aetiological cause and disease effect is less well understood.
ACKNOWLEDGMENTS We wish to thank Professor J.R. Pattison and his colleagues for their helpful collaboration in carrying out the DNA hybridisation and serological tests. REFERENCES Almeida JD, Waterson AP (1969): The morphology of virus-antibody interaction. Advances in Virus Research 15:307-338. Anderson MJ, Davis LR, Jones SE, Pattison JR, Serjeant GR (1982): The development and use of a n antibody capture radioimmuno-
assay for specific IgM to a human parvovirus-like agent. Journal of Hygiene (Cambridge) 88:309-324. Anderson MJ, Jones SE, Fisher-Hoch SP, Lewis E, Hall SM, Bartlett CLR, Cohen BJ, Mortimer PP, Pereira MS (1983): Human parvovirus, the cause of erythema infectiosum (fifth disease). Lancet 1:1378. Anderson MJ, Jones SE, Minson AC (1985): Diagnosis of human parvovirus infection by dot-blot hybridisation using cloned viral DNA. Journal of Medical Virology 15:163-172. Brown T, Anand A, Ritchie LD, Clewley J P , Reid TM (1984): Intrauterine parvovirus infection associated with hydrops fetalis. Lancet 2:1033-1034. Cohen BJ, Mortimer PP, Pereira MS (1983): Diagnostic assays with monoclonal antibodies for the human serum parvovirus-like virus (SPLV1. Journal of Hygiene (Cambridge1 9 1:113 -130. Cossart YE, Field AM, Cant B, Widdows D (1975): Parvovirus-like particles in human sera. Lancet 1:72-73. Kurtzman GJ, Ozowa K, Cohen B, Hanson GR, Oseas R, Young NS (1987): Chronic bone marrow failure due to persistent B19 parvovirus infection. New England Journal of Medicine 317:287294. Mitchell S, Welch J , Nicholson F, Weston-Smith S, Bradbeer CS (1990):Parvovirus infection and anaemia in a patient with AIDS: Case report. Genitourinary Medicine (in press). Pattison J R , Jones SE, Hodgson J, Davis LR, White JM, Stroud CE, Murtaza L (1981): Parvovirus infections and hypoplastic crisis in sickle-cell anaemia. Lancet 1:664-665. White DG, Woolf AD, Mortimer PP, Cohen BJ, Blake DR, Bacon PA (1985):Human parvovirus arthropathy. Lancet 1:419-421.
