Kaposi's sarcoma risk among transplant recipients in the United States (1993–2003)

Abstract

Kaposi's sarcoma (KS) risk is high in immunosuppressed transplant recipients. KS develops in recipients with pre-existing infection with human herpesvirus 8 (HHV-8), the causative agent for KS, but it can also occur in recipients infected by donors. The relative importance of these sources of infection in recipients in the United States is unknown. We report recipient and donor characteristics associated with KS among transplant recipients in the United States. KS risk, after solid organ transplantation during 1993–2003, was analyzed using data from the Organ Procurement and Tissue Network. Associations were determined using proportional hazards regression. Sixtyfive KS cases were identified among 234,127 transplants (incidence 8.8 per 100,000 person-years). Most cases occurred in the first 2 years after transplantation (incidence 12.5 per 100,000 person-years). KS risk increased steadily with recipient age (p_trend < 0.001) and was associated with the recipient being male (HR 1.8, 95% CI, 1.0–3.2), Hispanic (2.1, 1.1–3.8) and a non-U.S. citizen (3.9, 1.8–8.6). Mismatch at the HLA-B locus, but not at HLA-A or HLA-DR loci, was associated with heightened risk (HR 3.6, 95%CI 1.1–11 for 1–2 vs. 0 HLA-B mismatches). KS was unrelated to donor characteristics and was not significantly related to use of specific antirejection medications. Our study found that KS incidence was low among transplant recipients in the United States, but it was associated with recipient age, sex and citizenship, perhaps reflecting pre-existing HHV-8 infection. The high KS risk immediately posttransplant and in persons with HLA-B mismatch highlights the role of immunosuppression and/or immune stimulation in KS pathogenesis. © 2006 Wiley-Liss, Inc.

Kaposi's sarcoma (KS) is a soft tissue malignancy strongly linked to immunosuppression in solid organ transplant recipients1 and in persons infected with the human immunodeficiency virus (HIV).2 Human herpesvirus 8 (HHV-8, also called KS-associated herpesvirus) is considered a necessary, albeit, insufficient cause of KS.3 Among transplant recipients, KS develops mostly in those with pre-existing HHV-8 infection,4 and seropositive recipients have a 28–75-fold higher KS risk than do seronegative recipients.5, 6 However, KS can also develop in recipients who become infected via donated organs7, 8 and, as reported recently, after transmission of KS progenitor cells from donors.9 The contribution of each of these transmission-mechanisms to KS risk has not been determined but may vary according to regional HHV-8 seroprevalence.

KS risk in transplant-recipients parallels HHV-8 seroprevalence in different countries and regions. The cumulative risk of KS in recipients is highest in the Mediterranean and Middle East countries (2–6%),10, 11 where adult HHV-8 seroprevalence is also relatively high (10–30%).12 Data for sub-Saharan Africa, where HHV-8 seroprevalence is relatively high (35%),13 are sparse, but 1 study reported a risk of 0.6% per year among non-white recipients in South Africa.14 By contrast, the cumulative KS risk is low (0.4–0.6%) among transplant recipients in the United States,15 where HHV-8 seroprevalence is low (<3%),16 and the risk is considerably lower (<0.01%) among recipients in Japan17 and the Nordic countries.18 Postulated risk factors for KS in recipients include HLA mismatch of recipient to donor and specific HLA polymorphisms, but the evidence is inconsistent.6, 19 Immunosuppression due to drugs used to prevent rejection of organs has been consistently associated, but evidence for difference in risk according to the specific immunosuppressive drugs is not conclusive.20 Recently, a newly introduced antirejection medication, sirolimus, has been associated with regression of KS lesions in transplant recipients.21

Here, we investigate factors associated with KS among solid-organ transplant recipients in the United States, where the role of recipient and donor characteristics on KS risk has not been characterized. Characterizing epidemiological risk factors of KS in transplant recipients can shed light on KS pathogenesis and help inform decisions regarding transplantation policy.

Methods

We analyzed KS risk among transplants in the United States, using data from the Organ Procurement and Tissue Network (OPTN), the U.S. transplant network that includes all U.S. organ procurement organizations and transplant centers.22 OPTN activities include conducting recipient matching and organ placement, and collecting and managing scientific data about organ donation and transplantation. Transplant centers routinely provide recipient and donor clinical data to OPTN at transplantation, and follow-up data on recipients at 6 months after transplantation and yearly thereafter. Anonymized OPTN data are available for research use.

We obtained data on 316,607 transplants performed in the United States from 1987–2003. Data included baseline recipient and donor demographic characteristics (e.g., sex, age, race and citizenship), immunological variables (ABO and Rhesus blood type; HLA-A, B and DR mismatch scores) and HIV status. Using information from OPTN variables for race and country-of-origin, we identified recipients and donors born in the Middle East, Italy and Africa, areas where HHV-8 seroprevalence is relatively high. Data on initial medications used to induce and maintain immunosuppression, and on episodes of acute rejection at 6 months and 1 year after transplantation, were obtained. We also obtained data on the use of anti-herpesvirus drugs (acyclovir, ganciclovir and valganciclovir) within the first 12 months. Medication data specified the administration of particular medications, but not dosage. Finally, we identified KS diagnoses and their date of onset from clinical reports sent to OPTN by transplant centers. The method of reporting changed from a paper-based to a web-based system in 1999 and was associated with approximately a doubling in the annual number of reported KS cases by calendar year, which we address in our statistical analyses (see later).

We considered the transplant as the unit of analysis when estimating KS risk. Individuals were at risk for KS from the date of transplantation until the earliest of KS diagnosis, graft failure, retransplantation, death or last follow-up. Individuals who received multiple successive transplants contributed nonoverlapping time for each transplant. In the case of retransplantation, the succeeding transplant initiated a separate at-risk period. Few KS cases (n = 8 among 75,369 transplants) were reported to OPTN during 1987–1992, suggesting that KS ascertainment was incomplete in these early years. No KS cases were reported after a third or later transplant (n = 1,594 transplants) and in transplants performed in 2004 (n = 5,517), perhaps because of incomplete follow-up. To minimize underestimation of risk, transplants performed during 1987–1992 and 2004, and third or later transplants, were excluded from analysis.

KS incidence was estimated as the number of cases divided by the person-time at risk in the period, and the cumulative incidence was estimated using the Kaplan–Meier method. Hazards ratios (HRs) of KS across categories were calculated using Cox proportional hazards models. We included a time-dependent covariate in the models to control for possible changes in KS ascertainment by calendar year of diagnosis for periods of 1993–1998 and 1999–2003. To determine association of KS with acute rejection episodes at 6 months and 1 year, and use of anti-herpesvirus medications in the first year, we restricted analysis to follow-up time subsequent to those events. For specific analyses, subjects with missing data were excluded.

Results

Of 234,127 transplants performed during 1993–2003, 142,486 (61%) were in males and 91,641 (39%) in females. By race, 181,101 (77%) were in white, 40,945 (18%) were in black and 11,989 (5%) belonged to other racial groups. Most (216,437, 92%) were first transplants, and 17,690 (8%) were second transplants. The majority were U.S. citizens (226,202, 97%), and by ethnicity, most were non-Hispanic (200,029, 89%). There were 1,597 (0.7%) recipients from the Middle East, 90 from Italy and 64 from various countries in Africa. The median age at transplantation was 47 years (interquartile range [IQR] 35–56 years) and the median follow-up after transplantation was 2.7 years (IQR 0.9–5.0). HIV results were available for 101,489 (45%) recipients: 56,327 (24%) were negative or indeterminate, 45,092 (19%) did not disclose their status and 70 (0.03) were positive. No donor was HIV positive.

We identified 65 KS cases, of which 80% were only cutaneous and 20% involved viscera. This corresponds to an overall incidence of 8.8 per 100,000 person-years. The median time from transplantation to KS diagnosis was 1.49 years (IQR 0.68–2.28). KS incidence was highest in the first 2 years after transplantation (mean incidence 12.5 per 100,000 person-years), and declined thereafter (Fig. 1a). No KS cases were reported after 8 years posttransplant and none of the KS cases occurred in HIV positive recipients. The cumulative incidence was 0.06% at 8 years after transplant (Fig. 1b).

Figure 1. Kaposi's sarcoma incidence among transplant recipients in the United States. (a) KS incidence is shown (events per 100,000 person-years) in defined time intervals after transplant; the dashed horizontal line marks average KS incidence (12.5 per 100,000 person-years) in the 0–1.99 years after transplant. (b)–(f) Kaplan–Meier estimates of cumulative incidence for all subjects (b) are shown by sex (c), age group (d), citizenship (e) and HLA B mismatch (f). In these panels, the y-axis is truncated to better depict the small risks.

Associations of KS with recipient characteristics are shown in Table I. KS risk was higher in males than in females (HR 1.8, 95% CI 1.0–3.2; Fig. 1c) and increased with the recipient's age at transplantation (p_trend < 0.001; Fig. 1d). KS risk was marginally increased among blacks and other racial groups, compared with that among whites (Table I), and significantly increased among Hispanics compared with that among non-Hispanics (HR 2.1, 95% CI 1.1–3.8), and non-U.S. citizens compared with U.S. citizens (HR: 3.9; 95% CI 1.8–8.6; Fig. 1e). Although individuals from the Middle East comprised only 0.7% of recipients, they accounted for 8% of KS cases, including 5 of 7 KS cases among non-U.S. citizens. Only 1 KS case was observed in recipients from Italy, and no cases were reported among recipients from Africa.

Table I. Risk of Kaposi's Sarcoma Associated with Recipient Demographic Characteristics (1993–2003)

Characteristic	Person-years	KS cases	Hazard ratio1	95% CI	p-Value
KS, Kaposi's sarcoma; CI, Confidence interval. Missing data were excluded from analysis. 1 Hazard ratios were adjusted for calendar period of KS diagnosis (see Methods). * p-Value is test of heterogeneity. † †p-Value is test of trend.
Sex					0.03*
Female	290,686	17	Ref.
Male	451,059	48	1.8	1.0–3.2
Age (years)					<0.001†
<30	123,946	2	0.3	0.1–1.3
30–39	128,879	4	0.5	0.2–1.7
40–49	189,571	11	Ref.
50–59	184,045	24	2.2	1.1–4.4
>60	106,789	24	3.6	1.8–7.4
Year of transplant					0.86†
1993–1995	294,444	14	Ref.
1996–1998	254,306	23	1.2	0.5–2.7
1999–2001	161,673	24	1.1	0.4–3.4
2002–2003	31,322	4	0.8	0.2–3.8
Race					0.03*
White	586,781	44	Ref.
Black	118,573	15	1.6	0.9–2.9
Other	36,058	6	2.1	0.9–4.1
Ethnicity					0.02*
Non-Hispanic	619,398	50	Ref.
Hispanic	76,040	13	2.1	1.1–3.8
Citizenship					0.001*
United States citizen	718,354	58	Ref.
Non-United States citizen	21,465	7	3.9	1.8–8.6
Education					0.74†
Less than college	239,000	23	Ref.
College/higher level	177,047	22	1.3	0.7–2.3
Unknown	325,698	20	0.9	0.5–1.6
Organ type					0.66*
Kidney and/or pancreas	475,246	42	Ref.
Thoracic	117,742	13	1.3	0.7–2.5
Liver	148,749	10	0.8	0.4–1.5
Sequence of organ transplant					0.26*
First	696,463	63	Ref.
Second	45,283	2	0.4	0.1–1.8

Donor characteristics were unrelated to KS risk (Table II), except for donor citizenship. Compared with recipients of organs from U.S. donors, KS risk was increased in recipients of organs from non-U.S. donors (HR 2.4, 95% CI 1.1–5.3). However, recipient and donor citizenship were positively correlated (4% of U.S. citizens received organs from non-U.S. citizens vs. 22% of non-U.S. citizens). In a model including recipient and donor citizenship, recipient citizenship remained significantly associated with KS (HR 3.4, 95% CI 1.5–7.6), but the association with donor citizenship was attenuated (1.9, 0.8–4.4).

Table II. Risk of Kaposi's Sarcoma Associated with Donor Demographic Characteristics

Characteristic	Person-years	KS cases	Hazard ratio1	95% CI	p-Value
KS, Kaposi's sarcoma; CI, Confidence interval. Missing data were excluded from analysis. 1 Hazard ratios were adjusted for calendar period of KS diagnosis (see Methods). * p-Value is test of heterogeneity. † p-Value is test of trend.
Sex					0.11*
Female	308,093	34	Ref.
Male	433,643	31	0.7	0.4–1.1
Age (years)					0.34†
<30	320,399	24	0.7	0.4–1.4
30–39	147,320	11	0.7	0.3–1.5
40–49	142,790	16	Ref.
50–59	89,476	11	1.1	0.5–2.3
>60	36,501	3	0.7	0.2–2.5
Race					0.44*
White	636,880	52	Ref.
Black	84,264	12	1.7	0.9–3.2
Other	20,008	1	0.6	0.1–4.1
Ethnicity					0.23*
Non-Hispanic	637,028	55	Ref.
Hispanic	74,297	10	1.5	0.8–2.9
Citizenship					0.03*
United States citizen	705,388	58	Ref.
Non-United States citizen	33,650	7	2.4	1.1–5.3
Donor type					0.64*
Living-related	129,803	9	0.7	0.3–1.5
Living-unrelated	27,137	3	0.9	0.3–3.0
Cadaver	580,283	53	Ref.

KS risk increased with the number of mismatches at the HLA-B locus but not at the HLA-A or HLA-DR loci (Table III). KS risk was higher among recipients with 1-2 compared with 0 mismatches at the HLA-B locus (HR 3.6, 95% CI 1.1–11; Fig. 1f). Likewise, any mismatch at the HLA-A locus was associated with a borderline increased risk of KS (HR 2.2, 95% CI 0.9–5.1 for 1-2 vs. 0 mismatches). KS risk increased with increasing HLA mismatch composite score, derived from the 3 HLA loci (A, B and DR), but the trend was not significant (p = 0.10; Table III). KS risk was not related to recipient blood group (Table III).

Table III. Risk of Kaposi's Sarcoma Associated with Immunological Characteristics

Characteristic	Person-years	KS cases	Hazard ratio1	95% CI	p-Value
KS, Kaposi's sarcoma; CI, Confidence interval. Missing data were excluded from analysis. 1 Hazard ratios were adjusted for calendar period of KS diagnosis (see Methods). 2 Number of HLA alleles found in the donor that are not shared by the recipient. 3 Analysis was restricted to subjects remaining at risk for Kaposi's sarcoma after the episode of rejection in the first 6 or 12 months. * p-Value is test of heterogeneity. † p-Value is test of trend.
ABO blood group
A	296,157	28	Ref.		0.49*
AB	88,709	9	1.0	0.5–2.2
B	33,882	2	0.6	0.1–2.6
O	322,964	26	0.8	0.5–1.4
Rhesus blood group					0.92*
Negative	58,178	4	Ref.
Positive	398,147	42	1.5	0.5–4.3
Unknown	135,506	11	1.3	0.4–4.0
HLA-A mismatch2					0.64†
0	134,125	6	0.4	0.2–0.9
1	292,725	34	Ref.
2	241,907	18	0.6	0.4–1.1
HLA-B mismatch2					0.04†
0	110,333	3	0.3	0.1–0.9
1	268,718	25	Ref.
2	289,633	30	1.1	0.6–1.8
HLA-DR mismatch2					0.32†
0	158,505	9	0.8	0.4–1.7
1	311,315	29	Ref.
2	194,170	20	1.2	0.6–2.2
HLA mismatch2					0.10†
0	62,540	1	0.1	0.0–1.0
1	27,286	1	0.3	0.0–2.5
2	75,846	7	0.8	0.3–2.0
3	141,417	16	Ref.
4	145,565	11	0.7	0.3–1.4
5	140,692	15	0.9	0.4–1.9
6	69,322	7	0.8	0.3–2.1
Rejection in the first 6 months3					0.21*
No	506,523	59	Ref.
Yes	108,662	2	0.5	0.2–1.4
Rejection in the first year3					0.22*
No	387,305	54	Ref.
Yes	140,833	11	0.6	0.3–1.4

KS risk was not related to episodes of organ rejection in the first year of transplant versus later (Table III). The risk of KS did not vary according to the use of anti-herpesvirus drugs, although substantial data were missing (not shown). Risk also did not vary according to the use of specific drugs used to induce or maintain immunosuppression (Table IV), except for sirolimus, which was first licensed in 1999. Only 3,619 (1.7%) and 12,672 (6%) recipients had received this drug for induction and maintenance, respectively. Use of sirolimus for maintenance appeared to be associated with a decreased risk of KS (HR 0.1, 95% CI 0.0–1.4), but this result was based on only 3 cases and was not significant.

Table IV. Risk of Kaposi Sarcoma Associated with Initial Immunosuppressive Medications

Characteristic	Person-years	KS cases	Hazard ratio1	95% CI	p-Value*
KS, Kaposi's sarcoma; CI, Confidence interval. Missing data were excluded from analysis. 1 Hazard ratios were adjusted for calendar period of KS diagnosis (see Methods). 2 Antibody treatments include various polyclonal and monoclonal antibody preparations. 3 Data on sirolimus were missing for most recipients, because this drug was introduced only recently. * p-Values are for tests for heterogeneity.
Medications used for induction
Calcineurin inhibitors					0.66
None	489,640	42	Ref.
Tacrolimus	67,437	8	1.2	0.5–2.5
Cyclosporine	182,399	15	1.2	0.6–2.1
Both	2,268	0	–
Antimetabolites					0.43
None	478,191	41	Ref.
Azathioprine	167,532	11	1.1	0.5–2.1
Mycophenolate mofetil	91,021	13	1.3	0.7–2.5
Both	5001	0	–
Corticosteroids					0.86
No	228,116	19	Ref.
Yes	489,685	45	1.1	0.6–1.9
Antibody treatments2					0.22
None	587,007	56	Ref.
Yes	154,739	9	0.6	0.3–1.3
Sirolimus3					0.86
No	15,089	2	Ref.
Yes	6,300	1	1.2	0.1–13
Medications used for initial maintenance
Calcineurin inhibitors					0.49
None	62,884	4	Ref.
Tacrolimus	193,807	21	1.2	0.4–3.5
Cyclosporine	478,357	38	1.3	0.5–3.7
Both	6,696	2	4.2	0.8–23
Antimetabolites					0.96
None	180,546	15	Ref.
Azathioprine	292,083	18	1.0	0.5–2.2
Mycophenolate mofetil	262,494	31	1.1	0.6–2.0
Both	6,621	1	1.8	0.2–13
Corticosteroids					0.58
No	27,057	3	Ref.
Yes	690,744	61	0.7	0.2–2.3
Sirolimus					0.15
No	1,305	1	Ref.
Yes	20,084	2	0.1	0.0–1.4

Discussion

KS risk was significantly associated with recipient demographic characteristics (sex, age, race, ethnicity and citizenship). These associations parallel expected associations with HHV-8 prevalence23 and are consistent with KS arising from pre-existing HHV-8 infection in recipients. Studies conducted in Mediterranean Europe and the Middle East have demonstrated substantial risks of KS in transplant recipients with pre-existing infection.1, 6, 19 Interestingly, in our study, KS risk was highest in recipients born in the Middle East, perhaps reflecting the high HHV-8 seroprevalence among adults (10–30%) in their countries of birth.24, 25 Only 1 recipient from Italy developed KS, and none among African recipients, but the number of recipients from these groups was small.

We observed an extremely low incidence of KS among transplant recipients in the United States. As expected, the incidence was substantially lower than the incidence reported in recipients from Mediterranean Europe10, 11 and Africa.14 The low incidence may be due to the much lower HHV-8 prevalence in the U.S. general population16 compared with prevalence in populations in Mediterranean Europe10, 11 and Africa14 or partly due to under ascertainment, which we discuss further below. Our results mirror and complement those from the Cincinnati Transplant Tumor Registry (CTTR), which previously reported a high risk of KS in transplant recipients,15 and a predominance of KS in males and in individuals of Jewish or Mediterranean descent. The CTTR obtains data from ∼85 transplant centers in the United States and internationally. Of 144 KS cases reported to CTTR from 1968 to 1989 with data on nationality, 18 (12%) were from the United States.19 Similarly, of 368 cases reported to CTTR from 1982 to 2001, 63 (17%) were from the United States26 However, neither of those studies reported information on the size of the U.S. transplant population at risk, thus precluding calculation of incidence rates comparable to the rate we report.

Although low in absolute terms, the incidence of KS that we observed among recipients in the United States nonetheless represents a 54-fold higher risk compared with the general population, based on age- and sex-specific KS rates in the United States prior to the AIDS epidemic (Surveillance, Epidemiology and End Results data, not shown).27 This increase likely underestimates the relative risk of KS in transplant recipients, because the United States general population rates include a substantial number of transplant-related KS cases. Our results contrast with those from a previous study, conducted at the Princess Margaret Hospital in Canada in 1979, which described a 400–500-fold increase in KS risk in transplant recipients compared with a control group of the same ethnicity.28 Nonetheless, that estimate was based on only 4 cases.

Donor characteristics were not associated with KS, except for donor citizenship. However, donor and recipient citizenship were positively correlated, and the effect of donor citizenship was attenuated when we accounted for recipient citizenship. Only 5% of organ donors were non-U.S. citizens, and only a minority of those donors were from regions with high HHV-8 seroprevalence (470 donors [0.24%] were from the Middle East and 64 [0.03%] were from Africa), and so we may have failed to demonstrate an independent effect of donor citizenship. Although our results do not provide support for a major role of donor-determined KS risk among United States recipients, a smaller contribution cannot be excluded. Case reports from Europe have reported the development of KS in recipients who were infected by donors.7, 8

KS risk peaked in the 0–2-year period after transplantation. This finding mirrors results from Italy, where KS risk was 5-fold higher in the first year after transplantation than in subsequent periods.10, 29 The rapid development of KS implicates immunosuppression as a major risk factor in the immediate posttransplant period. Perhaps, immunosuppressed recipients fail to control HHV-8 infection, leading to uncontrolled viral replication and/or expansion of tumor progenitor cells, and progression to KS. The decline in KS incidence after 2 years posttransplant may be due to recovery of immunity as antirejection medications are tapered. Alternatively, the decline could also be due to exhaustion of KS susceptible individuals from the pool of HHV-8 infected recipients.30 It is conceivable that the decline could also be partly due to the use of antiviral drugs (e.g., ganciclovir or acyclovir). Our data were somewhat incomplete, but use of anti-herpesvirus medications within 12 months after transplantation appeared unrelated to KS risk. We did not observe an increased risk of KS following early rejection episodes, although rejection episodes should be a marker of recipients whose antirejection therapies would have been intensified resulting in deepening immunosuppression. We did not find significant associations of KS with specific immunosuppressive drugs, although previous studies have reported increased risk with use of polyclonal antibodies6 and cyclosporine-based regimens.31 Unfortunately, we could not take into account drug dosage, which may correlate better with intensity of immunosuppression. Our results are consistent with sirolimus having a protective effect on KS, but this result is based on small numbers and was not statistically significant.

KS risk was elevated in recipients who were mismatched at the HLA B and, to a lesser extent, at the A locus. Two prior studies have evaluated the association between KS and HLA mismatch, but failed to demonstrate increased risk.6, 19 Neither study provided detailed data on KS cases and controls to allow independent assessment. Our finding of elevated KS risk in recipients mismatched for class I major histocompatibity antigens highlights the role of impaired cellular immunity in KS pathogenesis. However, HLA-B mismatch could also be a marker for recipients who require more intense immunosuppressive antirejection therapies and/or for recipients prone to chronic immunostimulation. Immunostimulation probably increases KS risk by modulating release of cytokines (e.g., interleukin 2, interleukin 6 and VEGF32), which are central to development of KS.

An advantage of our study was the use of a large centralized database to study risk factors of KS, which is rare even in transplant recipients. One limitation was relying on routinely collected clinical data, which were not always complete. For example, baseline HHV-8 serostatus results were too incomplete for a meaningful analysis (not shown). Furthermore, KS ascertainment may have been incomplete because transplant centers may not diagnose or report all cases. Ideally, completeness of KS ascertainment could be improved by linkage to population cancer registries, but this was not done. Nonetheless, under-ascertainment of KS would likely be nondifferential, which should not invalidate our comparisons of KS risk across subgroups of recipients.

An important consideration arising from this study, but also raised by others,33, 34 is whether HHV-8 screening should be utilized to identify high-risk recipients or donors. HHV-8 screening could identify individuals with high anti-HHV-8 antibody titers or viremia, characteristics which have been associated with heightened risk of KS among HIV-infected individuals35 and transplant recipients.6, 7 Unfortunately, currently available HHV-8 serologic assays are not reliably sensitive and specific for screening in low HHV-8 prevalence populations.16 For the United States, an additional consideration is the low cumulative risk of KS that we observed, which needs to be balanced against the considerably higher risk of death if a recipient fails to get a timely graft.20 Carefully designed studies will be needed to inform policy, particularly in transplant centers caring for large numbers of recipients from high HHV-8 seroprevalence areas.

To conclude, our findings suggest that KS in recipients in the United States is associated with recipient characteristics, perhaps because of pre-existing HHV-8 infection. The rapid onset after transplantation and association with HLA mismatch highlight the role of cellular immunity or immunostimulation in the development of KS.

Acknowledgements

The authors thank Ms. Ruth Parsons and Mr. David Castenson at the Information Management Systems (Rockville, MD) for preparing analysis data files.