Demographics, Trends, and Cardiovascular Mortality in Kaposi Sarcoma Patients in the United States: An Analysis of Surveillance, Epidemiology, and End Results Database

ABSTRACT

Background

Kaposi sarcoma (KS) is a vascular neoplasm caused by human herpesvirus‐8. There is a paucity of data on the cause of death in patients with KS, especially the proportion of cardiovascular mortality. This study aimed to provide annual trends, demographic analysis, and cause of death analysis of KS in the United States.

Methods

The Surveillance, Epidemiology, and End Results (SEER) database was used to collect data from 2000 to 2020 from 17 registries.

Results

We identified 8247 patients with KS, with majority males (90.6%) and in the 40–59 age range (38.5%). More than half of the patients were single (53.8%). Hispanic patients constituted 23.9% of the population, while 23.1% were non‐Hispanic Black patients. Overall, the incident burden of KS decreased, but there was no significant change in the incident burden of KS among females. Furthermore, no significant change in incident burden was observed among the Hispanic and Black populations. Similarly, annual incident burden among single and never‐married individuals has been decreasing since 2014. Patients with KS and primarily cutaneous features (skin KS) have a favorable survival rate compared to those with visceral KS (p < 0.0001). Females, widowed, and Black/American Indian patients had a significantly worse prognosis (p < 0.0001). Married patients had better survival (p < 0.0001). Infectious causes constituted 49.8% of deaths, while 11.4% of deaths were cardiovascular‐related. Cardiovascular deaths were significantly more common in females and Hispanic patients (p < 0.001).

Conclusions

Our study provides significant insights into cardiovascular mortality among patients with KS and highlights annual trends, demographic distribution, and mortality rates.

1. Introduction

Kaposi sarcoma (KS) is a vascular neoplasm caused by human herpesvirus 8 (HHV‐8), which results in characteristic purple–brown lesions [1]. KS was originally described by Moritz Kaposi in 1872 [2]. This first depiction, most commonly seen in elderly males (ages 64–72 years) of Mediterranean, Eastern European, and South American origin, became known as the classic subtype [3]. Classic KS is characterized by pigmented cutaneous lesions localized to the lower extremities with limited mucosal or visceral involvement [2]. The endemic subtype, generally prevalent among children in sub‐Saharan Africa, displays an aggressive disease course with lymphadenopathy and visceral organ disease [2]. The iatrogenic subtype is observed in immunocompromised individuals, such as post‐solid‐organ transplant [2, 3]. The epidemic subtype is specific to patients with HIV and is an AIDS‐defining malignancy [2, 3]. A fifth subtype was identified following several reports of HIV‐seronegative KS in individuals with no immunosuppressive cause [3]. This subtype, which is now called nonepidemic KS, is seen in middle‐aged men who have sex with men (MSM) and presents in a fashion similar to classic KS, with predominant cutaneous involvement and an indolent disease course [3].

The underlying etiology common to all KS subtypes is the interplay between immunosuppression and HHV‐8 [2]. Geographically, the incidence of KS reflects the prevalence of HHV‐8; high rates of infection in sub‐Saharan Africa correspond with a high incidence of KS [2, 4]. While any form of immunosuppression can increase susceptibility to KS with HHV‐8 infection, HIV coinfection appears to be the strongest driver–disease incidence increases from 1 in 100,000 to 1 in 20 in HIV‐positive individuals [4, 5]. It is also noteworthy that in the United States, an area of relatively low HHV‐8 prevalence, 30%–60% of HHV‐8 is found in HIV‐positive MSMs, and 20%–30% in HIV‐negative MSM [6].

Data show that the incidence of epidemic KS has decreased dramatically following the introduction of antiretroviral therapy (ART) in 1996 [7, 8]. Mortality from epidemic KS decreased from 54% in 1980–1995 to 12.1% in 1996–2005 [8]. Interestingly, the incidence and mortality of classical KS remained stable during this time interval [8]. However, despite a favorable trend in most demographic subgroups, a paradoxical increase in the incidence among Black patients, specifically in the southern United States, has been reported [7, 9, 10]. Data also suggest an association between poverty and KS [9]. To ensure risk stratification, timely diagnosis, and appropriate treatment, it is crucial to bridge the gaps in the understanding of demographic trends in KS.

We aimed to identify demographic, socioeconomic status, and treatment factors related to KS incidence, analyze annual trends in burden and mortality, and specifically investigate the causes of mortality, particularly cardiovascular mortality, in patients with KS.

2. Materials and Methodology

The Surveillance, Epidemiology, and End Results (SEER) database, funded by the National Cancer Institute, is a comprehensive source of population‐based data. It gathers data on cancer incidence and survival from 17 population‐based cancer registries, covering ~26.5% of the United States' population. SEER*Stat software (Version 8.4.2) and the International Classification of Diseases Version 3 (ICD‐O‐3) were used to identify patients diagnosed with KS between January 1, 2000 and December 31, 2020. Patients diagnosed by autopsy, death certificates only, and/or nonmicroscopically confirmed methods were excluded from the study to ensure data accuracy and reliability. Demographic and clinical parameters, including age, sex, race, rural/urban continuum, marital status, median household income, treatment, year of diagnosis, and cause of death, were retrieved.

The primary study endpoint was overall survival. Secondary outcomes included factors associated with mortality. Categorical variables were presented as frequencies and percentages. Using disease burden in various racial subgroups and the proportion of various racial groups from the United States Census Bureau, 2020 (Hispanic, 16.3%; Black, 12.4%; White, 61.6%), population‐level race‐specific incidence if KS was determined.

Survival analysis using log‐rank analysis and Kaplan–Meier survival plots was performed to assess survival outcomes among patients with different characteristics. An initial univariate regression was performed to identify significant variables for inclusion in the subsequent multivariate regression model, aimed at identifying the variables associated with mortality. Hazard ratios (HR) and their corresponding 95% confidence intervals (CIs) were determined to quantify the associations. Trends by year were assessed by Mann–Kendall trend analysis of data. Kendall's τ statistic was used to assess the strength and direction of trends. Statistical Analysis System v9.4 was used for the analysis. Statistical significance was set at p < 0.05.

2.1. Ethical Statement

Ethical approval was not required for this study, as all data were obtained from publicly available, deidentified sources. Similarly, informed consent was not required since the data were deidentified and publicly accessible. The authors had no access to patient‐identifying information, and all analyzed data were fully anonymized. Data access occurred on August 14, 2024.

3. Results

3.1. Demographic Distribution

Of the 8247 patients with KS in our study, 7472 (90.6%) were male, and the remaining 775 (9.4%) were female. The majority of the patients, 3176 (38.5%), were in the 40–59 year age group. In terms of racial representation, the majority of patients were non‐Hispanic Whites with 3745 (45.4%) patients, followed by Hispanic (all races) patients at 1974 (23.9%) patients and non‐Hispanic Black patients, 1906 (23.1%). Non‐Hispanic American Indian or Alaska Natives had the lowest absolute burden; 54 (0.7%) patients. These findings are elaborated in Table 1.

Table 1.

Demographic and socioeconomic characteristics of 8247 patients with Kaposi sarcoma from the Surveillance, Epidemiology, and End Results (SEER) database, 2000–2020.

Variable (n = 8247)	n (%)
Age	0–19	14 (0.2%)
	20–39	2767 (33.6%)
	40–59	3176 (38.5%)
	60–79	1401 (17%)
	≥ 80	889 (10.8%)
Gender	Female	775 (9.4%)
	Male	7472 (90.6%)
Race	Non‐Hispanic White	3745 (45.4%)
	Hispanic (all races)	1974 (23.9%)
	Non‐Hispanic Black	1906 (23.1%)
	Non‐Hispanic Asian or Pacific Islander	305 (3.7%)
	Unknown	263 (3.2%)
	Non‐Hispanic American Indian or Alaska Native	54 (0.7%)
Rural/urban continuum code (data collection site)	Urban	7930 (96.2%)
	Rural	303 (3.7%)
	Unknown	14 (0.2%)
Marital status at diagnosis	Single (never married)	4438 (53.8%)
	Married (including common law)	1145 (17.5%)
	Divorced	306 (3.7%)
	Widowed	361 (4.4%)
	Separated	104 (1.3%)
	Unmarried or domestic partner	65 (0.8%)
	Unknown	1528 (18.5%)
Median household income (in USD per year)	≤ 54,999	677 (8.2%)
	55,000–74,999	3898 (47.3%)
	≥ 75,000	3668 (44.5%)
	Unknown	14 (0.2%)

3.2. Socioeconomic and Background Distribution

In terms of marital status, the largest group was single (never married; 53.8%), followed by married (including common law; 17.5%) individuals. Regarding urban and rural distribution, 96.2% of the patients were in urban settings and 3.7% in rural areas. The highest proportion of patients (47.3%) was in the $55,000–$74,999 median household income category, with only 8.2% having a household income of ≤ $54,999. In total, 3668 (44.5%) patients had a median household income of ≥ $75,000. These findings are summarized in Table 1.

The largest proportion of cases in our cohort comprised patients with cutaneous features of KS (skin KS, n = 6892, 83.6%), followed by those with visceral involvement (n = 407, 4.9%), with the most common viscera involved being the lungs (n = 70, 0.08%). Patients with visceral involvement had a significantly worse prognosis as compared to those with skin KS (p < 0.001).

When stratified by age, patients aged 60–79 years had more favorable survival outcomes compared with younger individuals. Age ≥ 80 years was associated with worse long‐term survival (p < 0.001) (Figure 1A). Female sex was associated with slightly lower long‐term survival than male sex, although no short‐term survival difference was noted (p < 0.001) (Figure 1B). KS with visceral involvement had worse survival than those with KS involving the skin (p < 0.001) (Figure 1C). Married individuals had a more favorable survival rate than widowed patients, who had a poor long‐term prognosis (p < 0.001) (Figure 1D). Patients with a higher income (≥ $75,000) had superior survival outcomes than those in the lower income bracket (< $75,000) (p < 0.001) (Figure 1E). Rural location was associated with a significantly lower survival rate than urban location (p < 0.001) (Figure 1F).

Figure 1.

Kaplan–Meier survival analysis of Kaposi sarcoma patients from the Surveillance, Epidemiology, and End Results (SEER) database from 2000 to 2020, stratified by age (A), gender (B), site (C), marital status (D), income (E), and housing (F).

The overall annual burden trends showed a significant and consistent decline, with minimal variability (p < 0.001, τ = −0.660) (Figure 2A). For both males and females, the annual burden significantly declined over the study period (males: p < 0.001, τ = −0.657; females: p = 0.006, τ = −0.439) (Figure 2B). Burden for primary skin involvement in KS declined significantly since 2000 (p < 0.001, τ = −0.740), while visceral cases did not significantly change (p = 0.077, τ = −0.292) (Figure 2C). The burden in the never‐married, widowed, and divorced population significantly decreased (p < 0.05 for each comparison), while it did not change significantly for married and separated individuals (p = 0.053 and p = 0.229, respectively), and only increased for individuals classified as being unmarried or in a domestic partnership (p < 0.001, τ = 0.698) (Figure 2D). In the < $75,000 per annum group, the annual disease burden increased from 2000 to 2013 with a subsequent decline, yet did not show statistical significance when analyzed over the entire study period (p = 0.904, τ = −0.024). For the > $75,000 per annum group, a significant decline was observed (p = 0.030, τ = −0.348) (Figure 2E). The annual trend in the urban setup showed a downward trend over a period of 20 years (p < 0.001, τ = −0.740), whereas in the rural settings, there was no significant change (p = 0.107, τ = −0.259) (Figure 2F).

Figure 2.

Graphs depicting annual trends in Kaposi Sarcoma from 2000 to 2020 from the Surveillance, Epidemiology, and End Results (SEER) database. Overall frequency (A), gender (B), site (C), marital status (D), income (E), and housing (F).

There were significant race‐based differences in survival rates. Hispanic, non‐Hispanic White, and Asian or Pacific Islander race groups had better overall survival than the Black and American Indians or Alaska Native groups (p < 0.0001) (Figure 3A). In terms of burden trends, non‐Hispanic White patients showed a consistent reduction in the annual incident burden from 2000 to 2020 (p < 0.001, τ = −0.826). Incident annual burden remained stable in the other groups (p > 0.05 for all comparisons) (Figure 3B).

Figure 3.

Survival and annual incidence trends of selected ethnicities with Kaposi sarcoma from the Surveillance, Epidemiology, and End Results (SEER) database from 2000 to 2020. Survival (A), and annual trend in incidence (B).

Mortality data from KS showed that 56.3% of diagnosed patients were alive. Among those who died, 49.8% died of infectious causes, including HIV; 11.4% died of a cardiac cause; and the remaining died of other causes (Figure 4).

Figure 4.

Cause of death in Kaposi sarcoma from the Surveillance, Epidemiology, and End Results (SEER) database from 2000 to 2020. Other causes include leukemias, complications from visceral sites, primary skin, self‐inflicted injury, etc.

Cox regression analysis for multivariate analysis identified advanced age (> 80 years) (HR 2.263; p = 0.033), non‐Hispanic Black race (HR 1.492; p = 0.001), and visceral involvement (HR 1.709; p = 0.001) as factors associated with poor prognosis (Table S1).

In the subgroup analysis, a significantly higher proportion of females (24.3%; n = 112) than males (9.5%; n = 297) died due to cardiac‐related causes (p < 0.001). There was a significant association between race and the cause of death (p < 0.001). The highest proportion of cardiac‐related deaths occurred in White patients (16.1%; n = 279), followed by Hispanic (10.4%; n = 79), American Indian or Alaska Natives (9.7%; n = 3), Asian or Pacific Islanders (5.9%; n = 7), and Black patients (4.1%; n = 38).

4. Discussion

The majority of patients in our study were in the 40–59 age range, male, single or never married, with non‐Hispanic Whites (45.4%) being the most common racial subgroup, followed by Hispanics (23.9%). While we observed an overall decline in KS diagnoses within our study, we believe this trend is likely a direct positive outcome of the widespread and effective use of ART among individuals with HIV. KS is a rare malignancy that occurs at highly elevated rates among persons with HIV. KS is also recognized as an AIDS‐defining condition and an indication of immune suppression [5]. Therefore, the absence of individual‐level HIV serostatus data in our study limits the interpretation of observed KS incidence and trend findings. Consequently, we cannot definitively determine whether the observed decrease in KS incidence reflects a true reduction in occurrence within the at‐risk population or an apparent decrease solely attributable to the widespread adoption of ART in people with HIV. There was no significant change in incident burden among females. Similarly, no significant change in incidence was observed among Hispanic and Black patients. KS involving the skin had a more favorable survival rate than visceral KS. Furthermore, married patients had better survival, whereas Black and American Indian patients with KS had a less favorable prognosis. Kumar and colleagues previously analyzed racial disparities in patients diagnosed with KS using the SEER database and demonstrated that 84.2% were non‐Hispanic Whites and 15.8% were African Americans diagnosed between 1973 and 2013 [11]. Factors such as unequal access to timely and consistent ART, often mediated by socioeconomic disadvantages including poverty, lack of insurance, and limited transportation, likely play a substantial role in these outcome disparities. African Americans constitute a disproportionately high percentage of new HIV diagnoses, and data indicate they have lower levels of HIV care and viral suppression, lower rates of receiving any HIV medical care, and lower adherence to HAART compared to non‐Hispanic Whites [11]. It is plausible that certain genetic or biological predispositions could exist within these populations; however, the evidence from prior research strongly suggests that social and structural factors are often the primary drivers of unequal health outcomes [11].

Regarding mortality, data suggest that 95.4% of HIV‐associated KS patients are male [12, 13]. Our study found males had a more favorable long‐term survival than females with KS. This is consistent with previous survival data in patients with HIV, where it has been reported that females have worse outcomes despite lower viral loads and higher CD4 counts [14]. Similarly, Luo and colleagues studied the rate of HIV‐associated KS in the United States from 2008 to 2016 and discovered that 95.6% were male and 4.4% were female, with 47.7% of the patients aged between 35 and 49 years [13]. These findings are likely rooted in the alternate immune response in females which tends to be prone to overactivity. Since the majority of KS cases in the USA are of the epidemic subtype, it is plausible that the phenomenon observed in HIV is the driving factor behind the disparity observed in KS. Our findings may also be partially explained by the systemic barriers to healthcare access that females sometimes face [15]. Females were also more likely to have lower income and higher unemployment rates, contributing to difficulties in accessing healthcare [15]. Single mothers may find it more difficult to find and afford childcare, in addition to the cost of medical treatment itself, creating significant barriers and delays in access to medical care.

Our study revealed that married individuals have improved survival rates compared with single patients. This finding supports a meta‐analysis that showed that married patients with a cancer diagnosis had a better overall survival compared to unmarried and widowed cancer patients [16]. Having a spouse is associated with greater emotional support, higher income, and better mental health, all of which can improve outcomes [17]. Additionally, our study found that urban‐dwelling patients had improved KS survival. A systematic review conducted by Afshar and colleagues also demonstrated decreased cancer survival in patients living in rural areas compared to those living in urban areas [16]. Patients from rural areas have greater travel distances to hospitals, and decreased patient education and health literacy [18]. Moreover, rural patients have increased socioeconomic disparities associated with increased physical inactivity, obesity, and alcohol use [16].

A study investigating KS in HIV‐seropositive patients from 2001 to 2009 found that visceral KS had a significantly higher mortality rate than cutaneous KS [18]. Studies investigating other vascular tumors, such as angiosarcoma, have also found that visceral disease had a worse prognosis than cutaneous disease [19]. Our study investigating KS follows this trend, although other studies investigating various aspects of the phenomenon have differing findings. One study examining gastrointestinal KS in AIDS patients found that the degree of immunosuppression, but not the presence of GI involvement, predicted mortality [20]. While the overall mortality due to cardiac causes in the USA is ~20%, age‐adjusted causes of mortality vary according to the age group [21]. The Centers for Disease Control documents that 6.6% of deaths in the 25–34 age range are due to cardiac causes, while in the 35–44‐year range, it is 14.7% [22], and the mortality for the 45–55 age range is 22.8% [22]. Our data set consisted mostly of patients in the 25–55 range with a further breakdown (Table S2). Cardiac mortality was more prevalent as age increased, whereas infection followed the opposite trend and decreased with age.

Although cardiac mortality in KS is lower than those in the general population, the overall mortality rate is higher. This is because of infection, which is a major cause of the overall mortality in KS. These patients may have concurrent disease with both infection and underlying cardiac disease from KS, although it is reasonable to assume that florid infection may be easier to identify as a cause of mortality and is coded more frequently, even in the setting of cardiac complications. If infectious causes do not influence the data, it is possible that the cause of death would more often be attributed to a cardiac etiology and would be reflected in the data.

Cardiac causes of death in patients with KS can be attributed to various factors. KS can involve the heart and lead to cardiac complications, such as pericardial effusion, myocardial infiltration, and valvular dysfunction [21]. These cardiac manifestations can contribute to heart failure and ultimately result in cardiac‐related mortalities.

AIDS‐related metastatic KS typically involves the visceral layer of the serous pericardium or the subepicardial fat. It often shows a predilection for the subepicardial adipose tissue adjacent to a major coronary artery, with or without involvement of the adventitia of the ascending aorta or pulmonary trunk. KS lesions in the pericardium can lead to pericardial effusion and subsequent life‐threatening cardiac tamponade [23]. It is also crucial to consider that cardiovascular issues in HIV patients can arise from other factors, including cardiotoxicity associated with ART and other medications. These drug‐related effects can manifest as coronary artery disease, cardiomyopathy, myocarditis, and various other adverse cardiac events. Therefore, the overall cardiac risk in this population is influenced by both the disease itself and its treatment [23].

Also, KSHV is a risk factor and indicator of poor prognosis in dilated cardiomyopathy [24]. Zhao and colleagues compared the KSHV DNA load and its miRNA, kshv‐miR‐K12‐1‐5p, between patients with dilated cardiomyopathy and those with nondilated cardiomyopathy [24]. They found that patients with dilated cardiomyopathy had increased levels of kshv‐miR‐K12‐1‐5p, KSHV seropositivity, and quantitative titers [24]. Additionally, the risk of death from cardiovascular causes or heart transplantation was higher in patients with both dilated cardiomyopathy and KSHV seropositivity [24]. Kshv‐miR‐K12‐1‐5p is released by cardiac endothelial cells in response to KSHV infection [24]. The miRNA then suppresses type I interferon signaling, which is involved in the expression of antiviral proteins, including MX1, OAS, and IRFs, and the production of cytokines. Thus, kshv‐miR‐K12‐1‐5p contributes to decreased viral clearance [24]. This leads to an increased risk of dilated cardiomyopathy and repeated infections with other cardiotropic viruses [24]. Interestingly, kshv‐miR‐K12‐1‐5p enhances the effects of coxsackievirus B3 [24]. Upon coxsackievirus B3 infection, interferon‐beta levels increased, but kshv‐miR‐K12‐1‐5p suppressed interferon signaling. Without interferon signaling for clearance, increased coxsackievirus B3 levels increase interferon‐beta levels [24].

Patients with HIV‐associated KS should be treated with ART [25]. Liposomal doxorubicin and paclitaxel are first‐line chemotherapy agents used for HIV‐associated KS [25].

5. Limitations

As with many database studies, our study has some limitations. First, the data do not reflect endemic populations and thus are not generalizable to this subtype. Second, our data do not encompass every region of the United States, which may have had different HIV rates. Third, our data did not include HIV‐positive or HIV‐negative serostatus to determine the distribution of the subtypes in our data set. Fourth, the database lacks information on key cardiovascular risk factors such as smoking status, diabetes, and hypertension, as well as other potentially relevant comorbid conditions. This prevents us from adjusting for these potential confounders in our analysis, thereby limiting our ability to establish a direct causal link between Kaposi's sarcoma and cardiovascular mortality. Lastly, the lack of treatment‐related information, such as the specifics of ART or chemotherapeutic agents administered, further limits our capacity to account for their potential independent effects on cardiovascular outcomes. It also restricts our ability to fully assess their impact on survival estimates.

6. Conclusions

KS is a complex disease with different forms and prevalence among sexes and races. Although the overall incidence of KS is decreasing, certain subpopulations, such as Black and Hispanic men, continue to be disproportionately affected. Treatment options for KS include local and systemic therapies as well as ART for HIV‐associated KS. Mortality rates among patients with HIV‐associated KS have improved with the use of ART; however, disparities remain. The cardiac causes of death in patients with KS are not entirely clear and are likely multifactorial, but further studies are needed for a detailed understanding. Additional research is necessary to better comprehend the biology and treatment of KS, particularly in populations with high incidence rates.

Author Contributions

Rozi Khan: conceptualization, investigation, data curation. Abdul Qahar Khan Yasinzai: writing – original draft, methodology, writing – review and editing, formal analysis, investigation, software. Aman Goyal: writing – original draft, writing – review and editing, methodology, conceptualization, formal analysis, investigation, software. Katharine Tracy: writing – original draft, writing – review and editing. Amir Humza Sohail: writing – original draft, writing – review and editing. Shalini Ballur: writing – original draft, writing – review and editing. Samuel L. Flesner: writing – original draft, methodology, writing – review and editing. Ahmed Mahmood: writing – original draft, writing – review and editing. Kalyani Ballur: writing – original draft, writing – review and editing. Surabhi Maheshwari: data curation, writing – review and editing, writing – original draft. Tamour Khan Tareen: writing – original draft, writing – review and editing. Mohamed Daoud: writing – original draft, writing – review and editing, project administration, resources. Pooja Sethi: writing – original draft, writing – review and editing. Thomas Alexander: writing – original draft, writing – review and editing, validation, visualization, data curation. Asad Ullah: writing – original draft, writing – review and editing, visualization, validation, conceptualization, data curation, supervision. All authors have read and approved the final version of the manuscript.

Ethics Statement

No ethical approval was required for this study, as all data were obtained from publicly available, anonymous, deidentified database and were therefore exempt from IRB approval.

Consent

No informed consent was required for this study, as all data were obtained from publicly available deidentified sources.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

The lead author Rozi Khan affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Supporting information

Supplemental Tables Kaposi Sarcoma.

Khan R, Yasinzai AQK, Goyal A, et al., “ Demographics, Trends, and Cardiovascular Mortality in Kaposi Sarcoma Patients in the United States: An Analysis of Surveillance, Epidemiology, and End Results Database,” Health Science Reports 8 (2025): 1– 9, 10.1002/hsr2.71186.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available within the article and its Supporting Materials. Dr. Mohamed Daoud had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis. All data are available from the corresponding author upon reasonable request.