Our cross-sectional study of 10,027 KS patients, 2004–2018, is the first and largest NCDB study and the largest analysis since the advent of highly active antiretroviral therapy (HAART) for HIV in 2008 (as per revised International AIDS Society-USA guidelines)9. We demonstrated a 6.4% increase in one-year survival for KS patients from 2004–2007 to 2016–2018 (p < 0.0001), which was likely due to significant advancements in HAART during our study period10,11, including wider adoption of novel drug agents, such as second-generation nucleoside analog reverse transcriptase inhibitors, (i.e. amdoxivir and elvucitabine), preventing many HIV + positive patients from progressing to AIDS12–14. Approval of non-nucleoside reverse transcriptase inhibitors, including etravirine, (2012)15, fusion inhibitors (2003), CCR5 inhibitors (2007), and integrase inhibitors (2013) may also have contributed to improved survival14.
We observed a higher average age at diagnosis of KS patients than studies performed prior to 2008. In a British prospective cohort study of 254 consecutive patients with histologically confirmed KS, 1996–2008, there was an overall lower mean age at diagnosis (39 years vs. 47.7 years, respectively), and lower proportion of women compared to our study (4% vs. 9.6%, respectively)16. Similarly, in Armstrong et al.’s retrospective Surveillance, Epidemiology, and End Results (SEER) database study of 12,066 US KS patients, 1975–2005, there were 3.7% males, with 7.3% of patients greater than 70 years old at diagnosis17. There has also been a trend towards increasing numbers of women diagnosed with KS over time. Therefore, the higher proportion of women with KS, taken together with advances in HAART therapy, prolonging life for HIV + patients18,19, may explain our finding of KS diagnosed at older ages.
We found that US Middle (21.8%), South Atlantic (21.4%), and Pacific (18.8%) regions had the highest KS prevalences. Similarly, a SEER study, 2000–2013, including only men under 55 years, demonstrated the greatest regional incidence ratios in the South (2.0 per 100,000 person-years) and West (2.4 per 100,000 person-years)20. Armstrong et al. demonstrated higher age-adjusted standardized incidence ratio of AIDS-related KS in metropolitan areas (i.e. San Francisco 7/100,00) compared to rural areas (i.e. Iowa and Utah 0.3/100,000)17. These studies, taken together, demonstrate significant differences in KS incidence by US geographic region and population density, which may be due to HIV being more prevalent in larger cities. For example, in the US, the prevalence of HIV was highest in San Francisco’s injection drug user population (14.3%) in 2004, coinciding with the first year of our study21.
Previous KS studies have suggested that lower income and uninsured patients have worse outcomes. In a retrospective study of 207 low and middle-income HIV + children and adolescents with KS in Malawai, 2006–2015, 7-year overall survival was only 37%, likely due to lack of access to newer more effective HAART therapies, and to chemotherapies other than bleomycin 22 . In single-center retrospective study of 191 US KS patients, patients without vs. those with Ryan White Assistance (an HIV/AIDS assistance program given to uninsured or under-insured patients) had significantly higher mortality (Hazard Ratio: 4.06) 23 . Similarly, in our cohort, one-year survival was approximately 10% greater in patients with private insurance compared to uninsured and Medicaid-insured patients. In addition, one-year survival was approximately 5% greater for patients earning $46,000 or more compared to those earning less than $30,000. Therefore, our study corroborates previous findings that low-income and Medicaid-insured or uninsured KS patients have worse outcomes.
The majority of patients in our cohort were White, the proportion of Black patients increased by 7.7% over the study period, and a greater proportion of Black vs. White patients were HIV+. A SEER database study of KS patients diagnosed between 1973–2013 showed that prior to HAART, the incidence of KS was proportionally greater in White patients compared to Blacks, however, the incidence rate among Whites declined 1989–1994, but not among Blacks24. The hazard ratio for Blacks vs. Whites increased from 1981–1995 (HR: 1.1) to 1996–2013 (HR: 1.6), reflecting worse relative mortality outcomes for Blacks. Blacks have had a slower rate of decline in HIV positivity with increasingly higher mortality than Whites20,25−27. Therefore, our observed trend of declining incidence of KS among Whites is likely due to the greater decline in HIV positivity among Whites compared to Blacks.
We found that when compared to HIV + patients, HIV- patients were older and more often female. In a single-center retrospective study of 20 HIV- KS patients, 1987–2009, there was a male predominance (75%), with an average age of 69.9 years at diagnosis28, similar to our HIV- cohort findings of male predominance (75% vs. 95.5% for HIV+) with an older age at diagnosis (70.1 years vs. 40.0 years for HIV+). Non-HIV immunosuppressed patients are at risk for developing KS, especially through the lymph node, gastrointestinal tract, and lung (almost rarely in the absence of cutaneous findings)29,30. In addition, the incidence of KS in patients with lymphohematopoietic malignancies is more than four times that of general population, highlighting that immunodeficiency (iatrogenic, malignant, or HIV-induced) is a risk factor for KS28,29. Therefore, we hypothesize that a proportion of our HIV- patients had systemic immunosuppression, which is more common in older age, contributing to development of KS. The greater mortality among HIV + vs. HIV- individuals was also associated with their greater CDCC score (an index of co-morbidities), lower income status, higher dependence on Medicaid or lack of insurance, presence of B symptoms, and presence of edematous or ulcerated lesions.
Only 964 (9.6%) of our overall cohort was female, underscoring the significant male-predominance of KS reported in previous studies16,31. Previous studies focusing on females with KS have been limited by small sample sizes, making characterization difficult. Furthermore, despite larger overall sample sizes in previous database studies spanning roughly three decades, we report the largest sample size of women with KS. We found that on average, females were older than males (65.1 vs. 45.8 years, respectively), likely due to the greater proportion of men who were HIV+ (70.8%) compared to women (31.6%). Men also had a greater number of co-morbidities, were of lower income status, more often Medicaid-insured and located in the Pacific US, more likely to be treated at an academic center, and more often report B symptoms with significantly fewer CD4 cells, markers of more advanced disease.
KS management has not been adequately studied in the SEER database largely because treatment information is absent for many patients and treatment-specific variables were not included in earlier database years17,20. We found that HIV + patients began treatment significantly later than HIV- individuals, possibly due to a greater proportion of HIV + patients being Medicaid-insured or in low-income groups, with less access to medical care. KS treatments include HAART, radiation therapy, surgery/cryosurgery, chemotherapy, and immunotherapy, with HAART used exclusively for HIV + individuals18,32,33. In our cohort, on average, surgery was performed approximately a week following diagnosis, consistent with National Comprehensive Cancer Network (NCCN) guidelines recommending surgery as first-line treatment for cutaneous symptomatic lesions, while HAART followed by surgery is first-line for asymptomatic individuals with AIDS-related KS34. Radiation treatment was given later than all other treatments (average time 75.2 days), likely because radiotherapy is typically used at later stages of the disease due to its significant side effect profile, in line with NCCN guidelines recommending that radiotherapy be reserved for relapsed or refractory therapy35–37.
This study is subject to several limitations, including its retrospective nature, and lack of cause-specific death in the database, making it challenging to determine whether mortality differences observed between various KS groups were due to other conditions. Treatment information did include efficacy. Although we attempted to capture the full spectrum of KS patients through this broad epidemiologic study, patients who did not present for initial care to NCDB-contributing cancer centers or patients who did not seek care, were not included, which would likely bias our survival percentages. While this is the most complete summary of KS data post-HAART, the relatively rare nature of the condition makes national analyses slightly lower-powered compared to other more prevalent skin cancers.