Download AIDS-Related Kaposi`s Sarcoma: Evaluation of - hem

AIDS-Related Kaposi’s Sarcoma: Evaluation of Potential New
Prognostic Factors and Assessment of the AIDS Clinical Trial
Group Staging System in the Haart Era—the Italian
Cooperative Group on AIDS and Tumors and the Italian
C o h o r t o f P a t i e n t s N a ı̈ v e F r o m A n t i r e t r o v i r a l s
By Guglielmo Nasti, Renato Talamini, Andrea Antinori, Ferdinando Martellotta, Gaia Jacchetti, Francesco Chiodo,
Giuseppe Ballardini, Laura Stoppini, Giovanni Di Perri, Maurizio Mena, Marcello Tavio, Emanuela Vaccher,
Antonella D’Arminio Monforte, and Umberto Tirelli
Purpose: To assess potential new prognostic factors and
to validate the AIDS Clinical Trials Group (ACTG) for AIDSrelated Kaposi’s sarcoma (AIDS-KS) staging system in the
highly active antiretroviral therapy (HAART) era.
Patients and Methods: We collected epidemiologic, clinical, staging, and survival data from 211 patients with AIDS-KS
enrolled in two prospective Italian human immunodeficiency
virus (HIV) cohort studies. We included in the analysis all
patients with the diagnosis of KS made from January 1996,
the time at which HAART became available in Italy.
Results: In the univariate analysis, survival was not influenced by sex, age, level of HIV viremia at KS diagnosis,
HAART at KS diagnosis (HAART-naı̈ve v HAART-experienced), or type of HAART combination. Regarding ACTG
classification, the 3-year survival rate was 85% for T0
patients and 69% for T1 patients (P ⴝ .007), 83% for S0
patients and 63% for S1 patients (P ⴝ .003), and 83% for I0
patients and 71% for I1 patients (P ⴝ .06). In the multivariate analysis, only the combination of poor tumor stage (T1)
and poor systemic disease (S1) risk identified patients with
unfavorable prognosis. The 3-year survival rate of patients
with T1S1 was 53%, which was significantly lower compared with the 3-year survival rates of patients with T0S0,
T1S0, and T0S1, which were 88%, 80%, and 81%, respectively (P ⴝ .0001).
Conclusion: In the era of HAART, a refinement of the
original ACTG staging system is needed. CD4 level does not
seem to provide prognostic information. Two different risk
categories are identified: a good risk (T0S0, T1S0, T0S1) and
a poor risk (T1S1).
J Clin Oncol 21:2876-2882. © 2003 by American
Society of Clinical Oncology.
K
use of HAART is associated with a favorable impact on epidemiologic characteristics and on the therapeutic management of KS.5-11
The suppression of HIV by HAART could influence various
factors that stimulate KS growth — directly, by causing a
decrease in the level of HIV Tat protein and inflammatory
cytokines, and indirectly, by improving immune function. Furthermore, protease inhibitors have been reported to have direct
antiangiogenetic activity,12 which is an important element in the
pathogenesis of KS. On the basis of these considerations, it is
likely that the development of KS during HAART has been
profoundly modified and that the natural history and prognostic
factors for survival have also changed.
Thus far, the effects of HAART on the natural history and
prognostic factors of KS have not been documented in sizable
cohort studies. Assessment of prognostic factors for AIDS-KS in
the era of HAART is a crucial issue, considering that so far, the
AIDS Clinical Trials Group (ACTG) staging system for
AIDS-KS has been used to provide accurate prognostic information on which therapeutic decisions and clinical trial planning
are based. The ACTG staging system for AIDS-KS was defined
in 1988 and classified patients into good- or poor-risk groups
based on tumor extent (T), immune system status (I), as
measured by CD4 T-lymphocyte count, and evidence for HIV1–associated systemic illness (S).13 Subsequently, Krown et al14
validated the proposed staging system showing that the ACTG
TIS classification effectively predicted survival of patients with
APOSI’S SARCOMA (KS), a potentially life-threatening
multifocal neoplasm, is the most common AIDS-associated malignancy.1 KS has been associated with AIDS since the
beginning of the epidemic, and the risk of KS has progressively
increased during the human immunodeficiency virus (HIV) epidemic until the mid-1990s when, coincident with the introduction of
highly active antiretroviral therapy (HAART), a sharp decline of KS
incidence was observed.2,3 Durable suppression of HIV replication
by HAART is associated with CD4 cell recovery and with a
decrease in morbidity and mortality rates due to opportunistic
infections and HIV-related cancers.4 In particular, the widespread
From the Division of Medical Oncology A, National Cancer Institute,
Aviano; Epidemiology Unit, National Cancer Institute, Aviano; Division of
Infectious Diseases, IRCCS Spallanzani, Rome; II Division of Infectious
Diseases, Sacco Hospital, Milan; Infectious Diseases Clinic, University of
Bologna, Bologna; Division of Infectious Diseases, Santa Croce Hospital,
Ravenna; Division of Infectious Diseases, S. Salvatore Hospital, Pesaro;
Infectious Diseases Department, University of Torino, Torino; Division of
Infectious Diseases, Cuggiono Hospital, Cuggiono; Infectious Diseases
Clinic, Sacco Hospital, Milan, Italy.
Submitted October 29, 2002; accepted May 5, 2003.
Address reprint requests to Umberto Tirelli, MD, National Cancer
Institute, Via Pedemontana Occidentale 12, 33081 Aviano (PN), Italy; email:
[email protected].
© 2003 by American Society of Clinical Oncology.
0732-183X/03/2115-2876/$20.00
2876
Journal of Clinical Oncology, Vol 21, No 15 (August 1), 2003: pp 2876-2882
DOI: 10.1200/JCO.2003.10.162
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2877
AIDS-RELATED KS IN THE HAART ERA
AIDS-KS. The study was conducted before the introduction of
HAART; data were collected from 281 patients within 34 ACTG
trial sites from April 1989 to January 1995. To date, no study has
focused on the assessment of prognostic factors since the
introduction of HAART, and, in particular, we do not know
whether the ACTG staging system for AIDS-KS is still appropriate and useful in predicting survival in the era of HAART.
To assess potential new prognostic factors and to validate the
ACTG staging system in the era of HAART, we collected
epidemiologic, clinical, staging, and survival data from 211
patients with AIDS-KS enrolled in two prospective Italian HIV
cohort studies since January 1996.
PATIENTS AND METHODS
Our analysis included data from two prospective cohort studies: the Italian
Cooperative Group on AIDS and Tumors (GICAT) and the Italian Cohort of
Patients Naı̈ve from Antiretrovirals (ICONA). GICAT has recruited 2,500
HIV-positive patients affected by cancer, from many centers throughout Italy
since 1986, and many of them have been enrolled in clinical trials. ICONA
is a multicenter observational study which has recruited 4,000 HIV-positive
patients naı̈ve for antiretrovirals at enrollment, from 65 infectious disease
centers in Italy since 1997. We included in the analysis all patients with an
initial diagnosis of KS made in January 1996 or later, the date at which
HAART became widely available in Italy, and the following criteria were
required: histologically confirmed KS, serologic evidence of HIV infection,
at least one follow-up visit after KS diagnosis, and treatment with HAART
before and/or after KS diagnosis. Two hundred eleven patients met the
inclusion criteria and entered the study. Epidemiologic and HIV-related
clinical data (age at time of KS diagnosis, sex, race, HIV exposure category,
HIV stage at KS diagnosis, immunologic and virologic information at KS
diagnosis, antiretroviral therapy history, vital status, eventual cause of death)
were available in the ICONA database and GICAT case report forms. A
questionnaire was sent to the ICONA and GICAT participating centers to collect
retrospectively the following additional information (data available for all
patients): KS staging (according to the ACTG criteria), visceral involvement,
diagnostic procedures used to assess visceral involvement, type of KS treatment,
and, when needed, follow-up update (data required for 139 patients).
Although the staging procedures were probably not uniform among the
various centers, the following criteria were required to define KS gastrointestinal (GI) tract involvement: positive endoscopic evaluation, and, when
endoscopy was not performed, oropharyngeal KS with signs and symptoms
of GI disease and without microbiologic evidence of GI tract infection. For
KS pulmonary involvement, the following criteria were required: positive
bronchoscopic evaluation or evidence of interstitial infiltrates on chest
radiograph with positive thallium scan and without microbiologic evidence
of pulmonary infection.
Patients were staged according to the ACTG criteria, which are based on
the evaluation of tumor extension (T), CD4 cell count (I), and patient’s
systemic status (S). Good tumor extension risk (T0) was defined as KS
confined to the skin and/or lymph-nodes and/or minimal oral cavity involvement and poor tumor extension risk (T1) as extensive oral disease, the
presence of tumor-associated edema or ulceration, or GI or other visceral
disease. Good immune system status risk (I0) was defined as CD4 cell count
greater than 200/␮L, and poor immune system status risk (I1) as CD4 cell
count ⱕ 200/␮L. Good systemic disease risk (S0) was defined as no history
of opportunistic infections, no B symptoms (unexplained night sweats or
fever, ⬎ 10% unexplained weight loss, or persistent diarrhea), and Karnofsky score ⱖ 70%; poor systemic disease risk (S1) was defined as a history
of opportunistic infection, B symptoms, other HIV-related illness, and
Karnofsky score lower than 70%.
Information on survival was obtained through an active follow-up on
verification of vital status of the patients up to April 15, 2002, and survival
was measured from the date of diagnosis to death. Survival analysis was
computed by the Kaplan-Meier method,15 and the log-rank test was used to
Table 1.
Patient Characteristics at Kaposi’s Sarcoma
Diagnosis
Characteristics
No. of Patients
Sex
Male
Female
Age, years
⬍35
35–44
ⱖ45
HIV-exposure category
Homosexual or bisexual
IVDU
Heterosexual
KS stage
Tumor extension
T0
T1
Immune system
I0
I1
Systemic disease
S0
S1
HIV-RNA viremia,
(copies/mL)
Undetectable
ⱕ30,000
⬎30,000
ART at KS diagnosis
Yes
No
190
21
87
68
55
107
36
52
75
136
60
148
86
125
12
48
111
51
160
Abbreviations: ART, antiretroviral therapy; IVDU, intravenous drug user; HIV, human immunodeficiency virus;
KS, Kaposi’s sarcoma.
test the difference between subgroups. Differences between subgroups were
also tested in univariate analysis using the Cox proportional hazards model
to compute the hazard ratio (HR) and corresponding 95% confidence interval
(CI).16 Covariates that were significant in the univariate analysis were also
tested in the multivariate model.16 As a final step, a Cox proportional hazards
model was fitted with interaction terms between some covariates. In all cases,
statistical significance was claimed for P ⱕ .05 (two-sided).
RESULTS
General Characteristics
The general characteristics of the 211 patients with KS are
presented in Table 1. One hundred fifty-nine patients were
recruited from the GICAT, and 52 from the ICONA. One
hundred ninety patients were male and 123 patients (59%) were
older than 35 years, with a median age at KS diagnosis of 37
years (range, 20 to 80 years). The majority of patients (52%)
were homosexual or bisexual, and 17% and 25% were drug users
and heterosexual, respectively. Fifty-one patients (24%) had
experienced an AIDS-defining disease before KS diagnosis.
Most patients had severe immune system impairment and uncontrolled HIV viremia at KS diagnosis, with a median CD4 cell
count of 86 cells/␮L (range, 0 to 1,117 ␮L) and median HIV
viremia of 89,000 copies/mL (range, ⬍50 to ⬎1,000,000 copies/
mL). Regarding KS staging, 136 patients had poor-risk tumor
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2878
NASTI ET AL
Table 2.
Univariate and Multivariate Analysis of Survival for Each of the Main Variables (T, I, S) in 211 Patients With AIDS-Related
Kaposi’s Sarcoma
Univariate*
Factor
Tumor extension
T0
T1
Immune㛳 system status
⬎100 CD4
ⱕ100
Systemic disease
S0
S1
Multivariate†
No. of Patients§
HR†
95% CI
␹12
P
HR†
95% CI
␹12
P
75
135
1
2.72
1.31 to 5.36
7.25
.007
1
2.59
1.25 to 5.38
6.52
.01
95
113
1
2.11
1.13 to 3.95
5.49
.02
1
1.61
0.84 to 3.10
2.03
.15
125
86
1
2.37
1.32 to 4.28
8.29
.004
1
2.10
1.14 to 3.88
5.61
.02
Abbreviations: HR, hazard ratio; CI, confidence interval.
* Values are determined by Cox proportional hazards model.
† Values are determined by Cox proportional hazards model including all factors reported in this table.
‡ All HRs of 1 indicate a reference category.
§ The sums may not add up to the total number of patients because of some missing values.
㛳 The original AIDS Clinical Trials Groups immune system cut point of 200 CD4 cells/␮L to distinguish good (CD4 ⱕ 200) and poor (CD4 ⬎
200) risk groups is not predictive of survival at the univariate analysis (P ⫽ .06).
extension (T1), 148 patients had less than 200 CD4 cells/␮L (I1),
and 125 patients had poor-risk systemic disease (S1). Only 17
patients had no poor-risk features (T0, I0, S0), whereas 48
patients had all of the poor-risk features (T1, I1, S1). Visceral
disease was present in 73 patients, with GI tract involvement in
51 patients (documented by endoscopy in all patients) and
pulmonary involvement in 31 patients (bronchoscopically confirmed in 20 patients). Sixty-nine patients received chemotherapy, with most of them being enrolled onto clinical trials. The
most frequent combination regimens were: paclitaxel-vinorelbine (33 patients), liposomal daunorubicin (19 patients), and
doxorubicin-bleomycin-vincristine (14 patients). Twenty patients underwent radiation therapy.
After a median follow-up of 24 months (range, 1 to 72 months),
median survival was not yet reached. The survival rate at 3 years
was 75%. KS progression was the most common cause of death,
representing 70% of deaths (33 patients), whereas opportunistic
infections were the cause of death in 30% of cases (14 patients).
Antiretroviral Therapy
All patients received HAART. Fifty-one patients were already
receiving therapy at KS diagnosis (31 patients were receiving a
triple combination therapy with protease inhibitors [PI], 14
patients were receiving a dual combination therapy with nucleoside reverse transcriptase inhibitors [NRTI], six patients were
receiving a triple combination therapy with nonnucleoside reverse transcriptase inhibitors [NNRTI]), and 160 patients initiated HAART at or after KS diagnosis (148 patients were
receiving a triple combination therapy with PI, and 12 patients
were receiving a triple combination therapy with NNRTI).
Among the 51 patients who were receiving antiretroviral therapies at KS diagnosis, 36 patients changed therapy and initiated a
novel HAART combination. Among them, 28 patients (all of the
14 patients on prior dual combination therapy with NRTI, all of
the six patients on prior triple combination therapy with NNRTI,
and eight patients on prior triple combination therapy with PI)
initiated a triple combination therapy with PI, and eight patients
(the eight patients on prior triple combination therapy with PI)
initiated a triple combination therapy with NNRTI. Fifteen
patients did not change their HAART regimen at KS diagnosis.
Prognostic Factors
In the univariate analysis, survival was not influenced by the
following variables: sex (HR ⫽ 0.86 for female v males; P ⫽
.79), age (HRs: 1.08 and 1.10 for 35 to 44 and ⱖ45 v ⬍ 35 years,
respectively; P ⫽ .90), level of HIV viremia at KS diagnosis
(HRs: 1.36 and 1.31 for ⬎30,000 to 100,000 and ⬎100,000 v ⱕ
30,000, respectively; P ⫽ .55), HAART (patients on triple
combination) at KS diagnosis (HR ⫽ 1.39 for HAART-experienced v HAART-naı̈ve; P ⫽ .36), and type of HAART combination (HR ⫽ 0.46 for triple combination without PI v triple
combination with PI; P ⫽ .45).
The correlation between survival and TIS variables is presented in Table 2. Poor-risk tumor extension (T1) and poor-risk
systemic disease (S1) were associated with significantly worse
survival, whereas poor-risk immune system status (I1) was not
associated with significantly shortened survival. The 3-year
survival rate was 85% for T0 patients and 69% for T1 patients
(P ⫽ .007), 83% for S0 patients and 63% for S1 patients (P ⫽
.003), and 83% for I0 patients and 71% for I1 patients (P ⫽ .06)
(Figs 1, 2, and 3). We investigated whether lower levels of CD4
cell count could better define prognostic groups and found that a
cut point of 100 CD4 cells provided the best discrimination.
Three-year survival rate was 82% for patients who had a CD4
count of more than 100, and 68% for patients with a CD4 count
of ⱕ 100 (P ⫽ .02) (Fig 4).
In the multivariate analysis, the Cox proportional hazards
regression model was fitted to evaluate the relationship between
TIS variables and survival. On the basis of the univariate
analysis, we used a CD4 cell count level cut-point of 100 to
discriminate good-risk (I0) and poor-risk (I1) immunologic
groups. In the multivariate model, only tumor stage (P ⫽ .01)
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2879
AIDS-RELATED KS IN THE HAART ERA
Fig 1. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
tumor extension status (T). T0, good tumor extension risk (——), 75 patients; T1,
poor tumor extension risk (- - -), 136 patients.
Fig 3. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
CD4 level. CD4 > 200 (——), 60 patients; CD4 < 200 (- - -) 148 patients.
and systemic disease (P ⫽ .02) remained independent predictors of
survival, whereas the CD4 count did not predict survival significantly. The increased hazard for patients with T1 (as compared with
T0) was 2.59 (95% confidence interval (CI), 1.25 to 5.38) and for
patients with S1 (as compared with S0) was 2.10 (95% CI, 1.14 to
3.88). To analyze the interaction between tumor stage (T) and
systemic disease (S) and its effect on survival, a Cox proportional
hazards model was fitted (Table 3). When considering patients with
good tumor and systemic disease stage (T0S0) at risk 1 (as reference
category), both subjects with good tumor extension and poor
systemic disease stage (T0S1) and subjects with poor tumor
extension and good systemic disease stage (T1S0) did not show a
significantly increased HR for death (HR ⫽ 1.70, 95% CI, 0.46 to
6.33; and HR ⫽ 2.14, 95% CI, 0.70 to 6.49; respectively). However,
when considering patients in whom both the tumor and systemic
disease stages were poor (T1S1), the HR was significantly increased
in comparison with those with T0S0 (HR ⫽ 5.68, 95% CI, 1.97 to
16.38), suggesting an additive interaction. Our analysis identifies
two different risk categories: a good-risk group (T0S0, T1S0, T0S1)
and a poor-risk group (T1S1) (Fig 5). The 3-year survival rate of
patients with T1S1 was 53%, significantly lower than the 3-year
survival rates of patients with T0S0, T1S0, and T0S1, which were
88%, 80%, and 81%, respectively (P ⫽ .0001). The median survival
for patients with T1S1 was 38 months, whereas the median survival
for patients with T0S0, T1S0, and T0S1 has not yet been reached.
Clinical practice and many reports in the literature indicate
that KS patients with pulmonary involvement show a particularly aggressive and life-threatening clinical course. For this
reason, we explored the possibility that patients with pulmonary
disease had a significantly worse survival compared with the
other patients with a poor-risk tumor stage disease, but without
pulmonary involvement. In the univariate analysis, the median
survival of the 31 patients with pulmonary disease was 26
months, while the median survival of patients without pulmonary
disease has not yet been reached. The 3-year survival rate of 46%
for patients with pulmonary involvement was significantly lower
than the 77% survival rate for patients without pulmonary
involvement (P ⫽ .0002) (Fig 6).
In light of this finding, we investigated the possibility that the
presence or the absence of pulmonary involvement (indicated in the
text as Tp1 and Tp0, respectively) and its interaction with the other
TIS variables would provide a better discrimination between risk
categories compared to the original ACTG tumor stage (T0, T1). In
the multivariate analysis, a Cox proportional hazards regression
model was fitted to evaluate pulmonary disease (Tp), immune
system status (I), and systemic disease (S) variables. Pulmonary
involvement (P ⫽ .001) and systemic disease (P ⫽ .03) were
Fig 2. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
systemic disease status (S). S0, good systemic disease risk (——), 125 patients; S1,
poor systemic disease risk (- - -), 86 patients.
Fig 4. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
immune system status (I); I0 > 100 CD4 (——), 95 patients; I1 < 100 CD4 (- - -),
113 patients.
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2880
NASTI ET AL
Table 3.
Combined Effects of Tumor Extension and Systemic Disease on Survival in 211 Patients With AIDS-Related Kaposi’s Sarcoma
Systemic Disease(s)*
S0
Factor
Tumor extension (T)
T0
T1
HR§
95% CI‡
S1
No. of Patients
HR§
95% CI
No. of Patients
HR
95% CI
HR§
95% CI†
45
80
1
2.14
1
—
—
0.70 to 6.49
30
56
1.70
5.68
2.44
0.46 to 6.33
1.97 to 16.38
1
2.80
—
—
—
1.35 to 5.80
—
—
1.35 to 4.39
Abbreviations: HR, hazard ratio; CI, confidence interval.
* HRs and 95% CIs based on the Cox proportional hazards model are adjusted for variables T and S.
† Adjusted for S variable.
‡ Adjusted for T variable.
§ HRs of 1 indicate a reference category.
significantly associated with poorer survival, whereas immune
system level of CD4 cells/␮L ⱕ 100 was not (P ⫽ .22). Table 4
shows the analysis of the interaction between Tp and S and its effect
on survival. Among patients without pulmonary involvement (Tp0),
S1 (Tp0S1) patients showed significantly increased hazard ratios for
death (HR ⫽ 2.68, 95% CI, 1.27 to 5.63) compared with S0
(Tp0S0) patients. Among patients with good-risk systemic disease
(S0), those patients with pulmonary involvement (Tp1S0) showed
significantly increased hazard ratios (HR ⫽ 4.98, 95% CI, 1.93 to
12.85) compared with patients without pulmonary involvement
(Tp0S0). Patients with both Tp1 and S1 evidenced the highest risk
for death, with an HR of 7.65 (95% CI, 3.24 to 18.04) compared
with Tp0S0 patients. This analysis identifies four risk categories with progressively increasing hazard ratios for death: Tp0S0
(HR ⫽ 1), Tp0S1 (HR ⫽ 2.68), Tp1S0 (HR ⫽ 4.98), and Tp1S1
(HR ⫽ 7.65).
DISCUSSION
The relationship between KS and HAART has been widely
investigated since the introduction of HAART in 1995, and
many aspects are being progressively clarified. The introduction
of HAART has altered the course of KS, and in particular has
significantly influenced epidemiologic, clinical, and therapeutic
aspects. The incidence of KS has dropped sharply since the
Fig 5. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
combined tumor extension (T) and systemic disease (S) risk categories. T0S0 (——),
45 patients; T0S1 (- - -), 30 patients; T1S0 (- - -), 80 patients; T1S1 (-䡠-䡠䡠䡠), 56
patients.
introduction of HAART,2,3 the use of HAART prolongs overall
survival of KS patients and is associated with an 80% reduced
risk of death among KS patients,17 KS responds to HAART in ⬎
50% of patients,5-7 and HAART is associated with prolonged
time to treatment failure in KS patients treated with systemic
treatments.18 There are several possible explanations for these
findings: the inhibition of HIV replication and the production
and release of the HIV-1 Tat protein, a KS progression factor,
the improvement of protective immune responses against human
herpesvirus 8 (HHV8), and the direct antiangiogenic effects of
some protease inhibitors.19,20
To date, however, there are no data in the literature which
evaluate potential new prognostic factors for survival in the era
of HAART, and, in particular, we do not know whether the
traditional ACTG staging system, based on tumor extent, immune system level, and systemic symptoms, still provides
correct and useful prognostic information in patients treated with
HAART. Krown et al have validated the ACTG staging classification as a valid predictor of survival in the pre-HAART era,
and showed that CD4 and tumor stage yield the most predictive
information. This study has been crucial for the management of
KS patients, because it provided useful information for clinical
and therapeutic decisions. In light of these considerations, we
have collected epidemiologic and clinical data on KS diagnosed
in 1996 or later within two large HIV cohort populations, with
Fig 6. Survival of patients with AIDS-related Kaposi’s sarcoma (AIDS-KS) by
pulmonary disease status. No pulmonary disease (- - -), 31 patients; Yes pulmonary disease (——), 105 patients.
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2881
AIDS-RELATED KS IN THE HAART ERA
Table 4.
Combined Effects of Pulmonary Disease and Systemic Disease on Survival: 211 Patients With AIDS-Related Kaposi’s Sarcoma
Systemic Disease(s)*
S0
Factor
Pulmonary disease
No
Yes
HR§
95% CI‡
S1
No. of Patients
HR§
95% CI
No. of Patients
HR§
95% CI
HR§
95% CI†
106
15
1
4.98
1
—
—
1.93 to 12.85
67
16
2.68
7.65
2.20
1.27 to 5.63
3.24 to 18.04
1
3.53
—
—
—
—
—
—
1.22 to 3.96
Abbreviations: HR, hazard ratio; CI, confidence interval.
* HRs and 95% CIs based on the Cox proportional hazards model are adjusted for the S and pulmonary disease variables.
† Adjusted for S variable.
‡ Adjusted for pulmonary disease variable.
§ HRs of 1 indicate a reference category.
the aim of assessing potential new prognostic factors and of
validating the ACTG staging system in the era of HAART. All
patients recruited in our study received HAART, either before or
after the KS diagnosis. We believe that the population is well suited
to the purpose of the study, since it comprises patients with different
baseline characteristics: those from the ICONA cohort with mainly
untreated and less advanced KS and those from the GICAT cohort
with advanced KS enrolled in chemotherapeutic trials. Furthermore,
all patients not enrolled in therapeutic trials entered the two cohorts
without being subjected to exclusion criteria, so that they represented a comprehensive KS population. A potential bias of our
study may be the fact that patients who never received HAART
were not included in the analysis, therefore potentially excluding
patients within the most favorable risk categories. However, we
think this is negligible because the majority of patients with KS
begin HAART independently of KS and HIV-related parameters. In
fact, only three patients were excluded from the analysis because
they never received HAART.
Unfortunately, information on HHV8 was not available in
our study. Effective biologic markers of KS could be a valid
tools in determining the prognosis of KS patients. Several
studies have indicated that HHV8 load is associated with KS
clinical stage and disease progression and is able to provide
clinically useful information.21,22
Analysis of our data indicated that sociodemographic factors
such as sex and age (in quinquennia), HIV-related factors (level
of HIV viremia), and HAART-related factors (HAART-naı̈ve v
HAART experienced at KS diagnosis, type of HAART combination) do not provide prognostic information. What our data
indicate is that in the post-HAART era, the factors included in
the ACTG staging classification that are most predictive for
survival have been modified. In the univariate analysis, neither
the CD4 cell cut-points of ⱕ 200 or ⱕ 150 predicted survival,
whereas tumor extension and HIV-related systemic disease
continued to provide significant prognostic information. We
explored lower CD4 cell count levels in order to define an
immune system cut-point predictive of survival and we found
that patients with CD4 cell counts ⱕ 100 had a significantly
poorer survival compared with patients with CD4 cell counts
greater than 100. However, the multivariate analysis indicated
that while tumor extension and systemic disease maintained their
correlation with survival, it excluded CD4 cell count above or
below 100 as predictive of survival. Importantly, the analysis of
the interaction between tumor stage and systemic disease and its
correlation with survival identified two main risk categories: the
group of patients presenting with both poor-risk tumor extension
and HIV-related systemic disease (T1S1) showed a significantly
increased risk of death compared with all the other groups
(T1S0, T0S1, T0S0), which all showed a similar longer survival.
Furthermore, survival analysis of patients with pulmonary involvement indicated that within the T1 risk category pulmonary
disease was associated with a significantly poorer survival
compared with the other T1 features.
These findings differ substantially from the pre-HAART
results of the Krown14 et al study, in which the CD4 count gave
independent predictive information, and tumor stage provided
additional predictive information in patients with a good immune
system status. We believe that HAART is probably responsible
for the alteration of the prognostic value of the ACTG classification in our study. The sociodemographic and clinical characteristics of our patients are similar to those of Krown et al
study.14 In analogy to our study, which recruited patients mostly
from the GICAT cohort in which half the patients had been
enrolled in clinical trials and treated with chemotherapy, the
study of Krown et al14 included all patients from ACTG
therapeutic trials in which nearly 50% of patients had been
treated with systemic chemotherapy. Furthermore, the distribution of risk categories was similar in the two studies, with most
patients in the poor-risk category for each of the TIS variables.
The finding that the level of immune deficiency does not provide
prognostic information in the era of HAART is not surprising.
The rapid immune reconstitution due to HAART may be a
reasonable explanation. Most cases of KS are the first AIDSdefining event and often are diagnosed in patients unaware of
their HIV seropositivity. Therefore, patients initiate HAART at
KS diagnosis and this leads in the majority of cases to a
significant immune restoration. The relative increase of CD4
lymphocyte counts after introduction of HAART has been
reported to have an independent prediction value for KS response.23 However, in our analysis we did not observe any
survival benefit in patients who began HAART after KS diagnosis compared to patients who were already receiving HAART.
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.
2882
NASTI ET AL
Of note, in our series, KS progression was the most common
cause of death (70% v 30% of patients who died of opportunistic
infections). Conversely, opportunistic infections were the most
common cause of death in the pre-HAART era, because of the
poor immune status that exposed KS patients to the infectious
complications of HIV more rapidly and significantly than to the
tumor progression itself. In the HAART era the increase of the
CD4 cell level is associated, in patients with KS, with a
markedly decreased risk of HIV-related infections and consequently KS survival is more influenced by T stage than by I
stage. The subgroup of patients with pulmonary involvement
showed the most unfavorable survival. Pulmonary involvement
is a sign of an aggressive clinical course and usually represents
late-stage disease, which is less likely to be influenced by
immune restoration. There are, however, some studies that report
prolonged survival in patients with pulmonary KS undergoing
chemotherapy and HAART.24
In the light of our results, we propose a refinement in the
application of AIDS-KS staging system, in which the immune
system should be eliminated as a prognostic determinant, and
only tumor extension and systemic disease should be considered as survival predictive variables. Two survival risk
categories are identified: poor risk (T1S1) and good risk
(T0S0, T1S0, T0S1). Furthermore, pulmonary involvement
predicts survival better than tumor extension and identifies the
poorest risk category, independent of the S variable. It is
noteworthy that survival analysis of the interaction between
pulmonary disease and systemic disease seems to provide a
better risk distribution between groups, with progressive HR
for death (Tp1S1⬎Tp1S0⬎Tp0S1⬎Tp0S0), as compared
with the interaction between classical tumor extension and
systemic disease. Further studies are clearly needed, in
particular with the aim of determining new prognostic factors,
which are more likely to be associated with KS (ie, HHV8)
than with HIV infection.
ACKNOWLEDGMENT
This study was supported by Istituto Superiore di Sanitá, Fondo Sanitarion
Nazionale, and Associazione Italiana per la Ricerca sul Cancro grants.
APPENDIX
The appendix is included in the full text version of this article only, available on-line at www.jco.org.
It is not included in the PDF version.
REFERENCES
1. Brodt HR, Kamps BS, Gute P: Changing incidence of AIDS-defining illness
in the era of antiretroviral combination therapy. AIDS 11:1731-1738, 1997
2. Jacobson LP, Yamishita TE, Detels R, et al: Impact of potent
antiretroviral therapy on the incidence of Kaposi’s sarcoma and nonHodgkin’s lymphomas among HIV-1-infected individuals. J Acquir Immune
Defic Syndr 21:S34-S41, 1999
3. Jones JL, Hanson DL, Dworkin MS, et al: Effect of antiretroviral
therapy on recent trends in selected cancers among HIV-infected persons. J
Acquir Immune Defic Syndr 21:S11-S17, 1999
4. Powderly WG, Landay A, Lederman MM: Recovery of the immune
system with antiretroviral therapy. The end of opportunism? JAMA 280:7277, 1998
5. Dupont C, Vasseur E, Beauchet A, et al: Long-term efficacy on
Kaposi’s sarcoma of highly active antiretroviral therapy in a cohort of
HIV-positive patients. AIDS 14:987-993, 2000
6. Vaccher E, di Gennaro G, Nasti G, et al: HAART is effective as
anti-Kaposi’s sarcoma therapy only after remission has been induced by
chemotherapy. J Acquir Immune Defic Syndr 22:407-408, 1999
7. Cattelan AM, Calabro ML, Aversa SML, et al: Regression of AIDSrelated Kaposi’s sarcoma following antiretroviral therapy with protease
inhibitors: Biological correlates of clinical outcome. Eur J Cancer 35:18091815, 1999
8. Dezube BJ: New therapies of the treatment of AIDS-related Kaposi’s
sarcoma. Curr Opin Oncol 12:445-449, 2000
9. Tavio M, Nasti G, Spina M, et al: Highly active antiretroviral therapy
in HIV-related Kaposi’s sarcoma. Ann Oncol 9:923, 1998
10. Lebbé C, Blum L, Pellet C, et al: Clinical and biological impact of
antiretroviral therapy with protease inhibitors on HIV-related Kaposi’s
sarcoma. AIDS 12:F45-F49, 1998
11. Nasti G, Errante D, Santarossa S, et al: A risk and benefit assessment
of treatment for AIDS-related Kaposi’s sarcoma. Drug Saf 20:403-425, 1999
12. Ensoli B, Stürzl M: Kaposi’s sarcoma: A result of the interplay among
inflammatory cytokines, angiogenic factors and viral agents. Cytokine
Growth Factor 9:63-83, 1998
13. Krown SE, Metroka C, Wernz JC: Kaposi’s sarcoma in the acquired
immune deficiency syndrome: A proposal for uniform evaluation, response
and staging criteria. J Clin Oncol 7:1201-1207, 1989
14. Krown SE, Testa MA, Huang J: AIDS-related Kaposi’s sarcoma:
Prospective validation of the AIDS Clinical Trials Group Staging Classification. J Clin Oncol 15:3085-3092, 1997
15. Kaplan EL, Meier P: Non-parametric estimation from incomplete
observations. J Am Stat Assoc 53:457-481, 1958
16. Cox DR: Regression model and life-tables. J R Stat Soc B34:178-220,
1972
17. Tam HK, Zhang ZF, Jacobson LP, et al: Effect of highly active
antiretroviral therapy on survival among HIV-infected men with Kaposi’s
sarcoma or non-Hodgkin’s lymphoma. Int J Cancer 98:916-922, 2002
18. Bower M, Fox P, Fife K, et al: Highly active anti-retroviral therapy
(HAART) prolongs time to treatment failure in Kaposi’s sarcoma. AIDS
13:2105-2111, 1999
19. Sgadari C, Barillari G, Toschi E, et al: HIV protease inhibitors are
potent anti-angiogenic molecules and promote regression of Kaposi’s sarcoma. Nat Med 8:225-232, 2002
20. Pati S, Pelser CB, Dufraine J, et al: Antitumorigenic effects of HIV
protease inhibitor ritonavir: Inhibition of Kaposi sarcoma. Blood 99:37713779, 2002
21. Quinlivan EB, Zhang C, Stewart PW, et al: Elevated virus loads of
Kaposi’s sarcoma-associated human herpesvirus 8 predict Kaposi’s disease
progression, but elevated levels of human immunodeficiency virus type 1 do
not. J Infect Dis 185:1736-1744, 2002
22. Pellet C, Chevret S, Francès C, et al: Prognostic value of quantitative
Kaposi sarcoma-associated herpesvirus load in posttransplatation Kaposi
sarcoma. J Infect Dis 186:110-113, 2002
23. Pellet C, Chevret S, Blum L, et al: Virologic and immunologic parameters that predict clinical response of AIDS-associated Kaposi’s sarcoma to highly
active antiretroviral therapy. J Invest Dermatol 117:858-863, 2001
24. Holkova B, Takeshita K, Cheng DM, et al: Effect of highly active
antiretroviral therapy on survival in patients with AIDS-associated pulmonary
Kaposi’s sarcoma treated with chemotherapy. J Clin Oncol 19:3848-3851, 2001
Downloaded from jco.ascopubs.org on June 19, 2007 . For personal use only. No other uses without permission.
Copyright © 2003 by the American Society of Clinical Oncology. All rights reserved.