Classical_Kaposi_sarcoma

Text Material Preview

Classical Kaposi Sarcoma
Prognostic Factor Analysis of 248 Patients
Baruch Brenner, M.D.
1,2
Alina Weissmann-Brenner, M.D.
1,2
Erica Rakowsky, M.D.
1,2
Sara Weltfriend, M.D.
3,4
Eyal Fenig, M.D.
1,2
Rachel Friedman-Birnbaum, M.D.
3,4
Aaron Sulkes, M.D.
1,2
Shai Linn, M.D.
4,5
1 Institute of Oncology, Rabin Medical Center, Bei-
linson Campus, Petah Tiqva, Israel.
2 Sackler Faculty of Medicine, Tel Aviv University,
Tel Aviv, Israel.
3 Department of Dermatology, Rambam Medical
Center, Haifa, Israel.
4 Technion-Israel Institute of Technology, Haifa,
Israel.
5 Department of Epidemiology, Rambam Medical
Center, Haifa, Israel.
Address for reprints: Baruch Brenner, M.D., Insti-
tute of Oncology, Rabin Medical Center, Beilinson
Campus, Petah Tiqva 49100, Israel; Fax: 011-937-
3-9242087; E-mail: brennerb@mskcc.org
Received February 11, 2002; revision received
May 28, 2002; accepted June 10, 2002.
BACKGROUND. Classical Kaposi sarcoma (CKS) is a rare indolent neoplasm that is
particularly prevalent among Jews of Ashkenazi and Mediterranean origin. Data
regarding prognostic factors for CKS are scarce. The aim of the current retrospec-
tive analysis was to better define prognostic subgroups among patients with CKS.
METHODS. Between 1960 and 1995, 248 consecutive patients with CKS were treated
at the Rambam and Rabin Medical Centers in Israel. Although treatment options
included local excision, radiotherapy, and chemotherapy, observation alone was
used for 31% of patients. For prognostic factor analysis, disease progression was
classified as any progression and dissemination, and progression-free survival was
calculated for each.
RESULTS. At a median follow-up of 20 months, four patients (1.6%) died of CKS. Of
the patients eligible for analysis, 94 of 220 (39%) had any progression and 23 of 120
(18%) had dissemination. Only 8 of 202 (4%) had visceral spread. On univariate
analysis, age was a statistically significant prognostic factor for any progression (P
� 0.04), whereas immunosuppression and visceral involvement at presentation
had only borderline significance. Immunosuppression was the only prognostic
factor for dissemination (P � 0.003). On multivariate analysis, both age and
immunosuppression were significant prognostic factors for any progression (P
� 0.001 and 0.01, respectively). Immunosuppression was also predictive of dis-
semination (P � 0.006).
CONCLUSIONS. Immunosuppression and older age (50 years and older) are strongly
associated with poorer outcome among CKS patients. The two end points used in
this study may be used for future prognostic factor analyses. Cancer 2002;95:
1982–7. © 2002 American Cancer Society.
DOI 10.1002/cncr.10907
KEYWORDS: classical Kaposi sarcoma, prognostic factors, immunosuppression,
older age.
C lassical Kaposi sarcoma (CKS) is a rare neoplasm, first described
by Moritz Kaposi in 1872 as an idiopathic, multipigmented sar-
coma of the skin. Its predisposition among Jews of Ashkenazi and
Mediterranean origin is well recognized, as is its prevalence among
men.1 The course of CKS is usually indolent over many years and is
generally not life threatening, as opposed to the aggressive course
associated with the African or the acquired immunodeficiency syn-
drome (AIDS)-related Kaposi sarcoma (KS).2 The morphology is sim-
ilar among all forms KS and is characterized by endothelial-linked
vascular channels, spindle-shaped cells, and hemosiderin-containing
macrophages.3
The etiology of KS is unknown. However, several epidemiologic
and environmental factors, as well as immunosuppression, play a role
in the development and clinical course of the disease. In the last
1982
© 2002 American Cancer Society
decade, epidemiologic and biologic evidence has sug-
gested that a recently discovered herpes virus, named
Kaposi sarcoma herpes virus or human herpes virus 8
(HHV8), is a required infectious cofactor responsible
for all known forms of KS.4 Clinical management may
consist of radiotherapy, chemotherapy, excision of le-
sions, or observation alone.5–7
Data regarding prognostic factors for CKS, as well
as for the other forms of KS, have been scarce. This
represents a major drawback in allowing the adjust-
ment of the appropriate treatment approach to the
individual patient. We present a retrospective analysis
of a large number of CKS patients from two of the
largest medical institutions in Israel, the Rambam and
the Rabin Medical Centers. The aim of the current
study was to identify the prognostic factors that influ-
ence the clinical course of the disease.
MATERIALS AND METHODS
Patients
Between 1960 and 1995, 248 consecutive patients with
CKS were treated at the Rambam and Rabin Medical
Centers (125 and 123 patients, respectively). The
former serves most of the population in northern Is-
rael, whereas the latter is located in the greater Tel
Aviv Metropolitan area. Diagnosis was confirmed in all
cases by histopathologic examination. Upon presen-
tation, patients underwent physical examination and
routine hematologic and chemistry analysis. Serologic
screening for human immunodeficiency virus, nega-
tive in all cases, has been performed since 1984. Fol-
low-up was based essentially on history and physical
examination. Diagnostic interventions, including var-
ious imaging tests and laparoscopic procedures, were
applied as appropriate.
Data were collected from the patients’ medical
files and the following parameters were recorded: gen-
der, origin, age at diagnosis, anatomic distribution,
immunocompromised conditions, second malignan-
cies, clinical course, and treatment modalities. Patient
characteristics are summarized in Table 1. The male-
to-female ratio was 2.3:1. The median age of onset
among CKS patients was 65 years (range, 20 –92 years).
Of 248 patients, 47 (19%) had second malignancies,
most frequently occurring adenocarcinomas and lym-
phoreticular disorders, and 31 (12%) had an underly-
ing immunocompromised state.
Clinical Presentation and Treatment
Clinical features at presentation are summarized in
Table 2. The limbs were involved at diagnosis of CKS
in 202 patients (81%). Disease distribution was defined
as limited when lesions could be treated by local mea-
sures, such as surgery or radiotherapy, or diffuse when
the extent of disease necessitated systemic treatment.
Anatomic distribution was categorized as limited and
diffuse acral, truncal with or without spread to mu-
cous membranes, and visceral.
Treatment options included local excision, radio-
therapy, and chemotherapy. More than one treatment
modality was used in 80 patients. It is noteworthy that
observation alone was the treatment approach in 76
(31%) patients (Table 2). Progression of disease was
classified into two categories: 1) any progression: any
TABLE 1
Patient Characteristics
No. of patients (%)
Gender
Male 173 (70)
Female 75 (30)
Origin
Ashkenazi Jew 146 (59)
Sephardi Jew 96 (39)
Arab 6 (2)
Second malignancies
Solid tumor 24 (10)
Lymphoreticular 15 (6)
Skin tumora 8 (3)
Immunocompromised state
Organ transplantation 10 (4)
Steroid therapy 19 (8)
Renal failure 2 (1)
a Basal and squamous cell carcinomas of skin only.
TABLE 2
Clinical Features
No. of patients (%)
Primary site
One limb 94 (38)
Two to four limbs 108 (43)
Trunk 33 (13)
Mucous membranes 11 (5)
Visceral 2 (1)
No. of lesions at presentation
Single 66 (26)
Multiple 182 (74)
Distribution at presentationa
Limited 163 (66)
Diffuseb 85 (34)
Treatment approachc
Local excision 90 (36)
Radiotherapy 141 (57)
Systematic chemotherapy 48 (19)
Observation alone 76 (31)
a See definitions in Materials and Methods.
b Includes two patients with visceral disease.
c More than one treatment modality was used in 80 patients.
Classical Kaposi Sarcoma/Brenner et al. 1983
local or diffuse spread of lesions, as well as evidence of
any development requiring initiation or change in
treatment, such as bleedingor pain; 2) dissemination:
progression from limited to diffuse disease. Patients
with diffuse disease at presentation were excluded
from this analysis.
Statistical Analysis
Because of the indolent nature of CKS, we chose pro-
gression-free survival (PFS) and not overall survival as
the main end point. A minimum follow-up of 1 month
was required for inclusion of patients in the PFS anal-
ysis. PFS was calculated for each of the two categories
of disease progression and it was defined as the inter-
val between the date of diagnosis and the date of
documentation of any of the events. The influence of
the following epidemiologic and clinical variables on
the course of CKS was estimated by the Kaplan–Meier
product limit method:8 age, gender, ethnic origin, ex-
istence of a second primary, immunosuppression, an-
atomic distribution, number of lesions, and treatment
modality. The anatomic distribution was categorized
as limited acral (61%), diffuse acral (19%), truncal or
mucosal (19%), or visceral (1%), according to the lo-
cation and the status (limited or diffuse) of the lesions
at presentation. Cox proportional hazards regression
models9 were applied for multivariate analysis. Risk
ratios and 95% confidence intervals were calculated
from the models. Statistical analysis was performed
using SPSS software.
RESULTS
Clinical Course
With a median follow-up of 20 months (range, 1–360
months), four patients (1.6%) died of CKS. Of the 220
patients with a follow-up equal to or longer than 1
month, who were thereby eligible for the prognostic
factor analysis, 94 (39%) had any progression of CKS
lesions, with a median time to progression of 15.4
months (range, 1–276 months). Dissemination of dis-
ease occurred in 23 of the 130 patients (18%) who were
eligible for this analysis, with a median time of 22.4
months (range, 2–156 months). Visceral spread was
evident in 8 of the 202 eligible patients (4%), with a
median PFS of 33 months (range, 2–192 months).
Univariate Analysis
Univariate analysis is shown in Table 3. With regard to
any progression, being younger than 50 years old was
a statistically significant and favorable prognostic fac-
tor. As illustrated in Figure 1, patients who were
younger than 50 years had a 75% 3-year any PFS.
Patients aged 50 years or more had a 55% 3-year any
PFS (P � 0.04). An immunosuppressive state (organ
transplantation, steroid treatment, or renal failure)
and visceral involvement at presentation were associ-
ated with poorer outcome, but were not statistically
significant (P � 0.06 and 0.07, respectively). Gender,
ethnic origin, second primary tumor, number of le-
sions, and initial management (any treatment vs. ob-
servation only) did not influence the occurrence of
any progression.
When the same analysis was applied for dissemi-
nation of CKS, immunosuppression was the only fac-
tor that predicted outcome, this time with distinct
statistical power (P � 0.003). Patients suffering from
an immunocompromised state had a 36% 3-year dis-
semination-free survival, whereas immunocompetent
patients had an 89% 3-year dissemination-free sur-
vival (Fig. 2).
Multivariate Analysis
Of the various multivariate analysis models for occur-
rence of any progression that were performed (data
not shown), the one depicted in Table 4 was chosen
due to its clinical relevance and high statistical power.
According to this model, both age and immunosup-
pression influenced PFS in a statistically significant
manner (P � 0.001 and 0.01, respectively). Multivari-
ate analysis for dissemination is shown in Table 5. For
this category of progression, only immunosuppression
was a statistically significant prognostic factor (P
� 0.006).
DISCUSSION
In the context of the typical indolent course of CKS,
assessment of the actual risk of either exacerbation or
dissemination of the lesions represents a real chal-
lenge. Addressing this issue may play a key role in both
the evaluation of the prognosis of the individual pa-
TABLE 3
Univariate Analysis: Statistically Significant Variables
Variables
No. of
patients
Free of any
progression (%) P value
Occurrence of any progressiona
Age (yrs)
� 50 17 71
� 50 203 56 0.04
Disseminationb of CKS lesions
Immunosuppression
Trans/steroids/RF 10 60
None 131 85 0.003
CKS: Classical Kaposi Sarcoma; Trans: organ transplantation; RF: renal failure.
a Any local or diffuse spread of lesions, any evidence of any development requiring initiation or change
in treatment.
b Spread of CKS lesions from limited to diffuse disease.
1984 CANCER November 1, 2002 / Volume 95 / Number 9
tient and the determination of the appropriate treat-
ment approach. Currently, clinicians are not guided
by any specific factors known to have an established
influence on the course of CKS. The relatively high
prevalence of the disease in Israel enabled us to study
a considerably large group of patients, allowing for an
effective prognostic factor analysis. Such analyses
could classify patients with CKS into separate catego-
ries, for each of whom the most adequate treatment
approach could be tailored.
Most of the demographic and clinical features of
our study group are in accordance with previous find-
ings in the literature. The 2.3:1 male-to-female ratio
found in our study has been reported previously,1,10 as
is the fact that most patients presented with CKS in
the seventh and eighth decades of their lives.11–13 Sim-
ilarly, the finding that 19% of our patients had coex-
isting malignancies is not unusual.10,14 –16 Twenty-
eight percent of our patients had an underlying
immunocompromised condition, which is also in ac-
cordance with previous reports.17,18
To perform the prognostic factor analysis, we had
to overcome two unique problems presented by CKS
patients. First, there are no features that are widely
accepted as associated with a more aggressive course
of the disease, as opposed to the AIDS-related KS, a
fact that is expressed by the absence of any staging
system, or equivalent. Second, the characteristically
indolent course of the disease makes standard out-
come measures, such as overall and disease-free sur-
vival, grossly irrelevant. Therefore, we had to define
specific parameters for disease spread at presentation,
as well as adequate end points. Regarding the latter
issue, we made a distinction between two categories of
disease progression: occurrence of any progression
FIGURE 1. Progression by age. (A) Any progression. (B) Dissemination. FIGURE 2. Progression by immunosuppression. (A) Any progression. (B)
Dissemination.
Classical Kaposi Sarcoma/Brenner et al. 1985
and dissemination. The assumption that these two
categories represent escalating clinical situations, as
do disease-free survival and distant disease-free sur-
vival among patients with other malignancies, is sup-
ported by several observations. First, there is an in-
creasing database regarding the clinical behavior of
CKS, which has generated earlier efforts to define dis-
ease progression.11,19 Second, the fact that involve-
ment of sites other than extremities, especially vis-
cerae, is characteristic of the more rapidly progressing
AIDS-related subtype of KS, suggests that these fea-
tures are associated with a more aggressive entity.
Third, in our study, we found an inverse correlation
between the severity of the event and the number of
affected patients (39% and 18% for any progression
and dissemination, respectively). Moreover, the me-
dian time for development of the events defined as
more severe was longer (15 and 22 months, respec-
tively). This pattern suggests a process of evolving
circumstances.
In the current study, univariate analysis identified
two unfavorable prognostic factors, age 50 years or
older at diagnosis and the existence of an immuno-
compromised state. When various multivariate analy-
sis models were applied, these variables constantly
retained theirstatistical significance in predicting oc-
currence of any progression. The latter variable also
strongly predicted dissemination of disease.
Immunocompromised patients are prone to de-
velop KS.17,18,20 Moreover, the association between the
aggressive clinical course of KS and immunosuppres-
sion has previously been suggested. In their study of
850 kidney transplant patients, Toth et al.21 described
two patients with KS were identified, both of whom
experienced an aggressive course of disease. Similarly,
Klepp et al.17 reported a disseminated and aggressive
course in three of seven patients who developed KS
during treatment with corticosteroids or radiotherapy.
Haim et al.22 observed an exceptionally virulent clin-
ical course in one of two patients who developed KS
during immunosuppressive treatment. Some authors
have described fluctuations in the severity of KS cor-
relating with the patient’s immune state. Most of these
reports are derived from the context of immunosup-
pressive treatment in kidney transplant patients.17,18,21
In these patients, KS lesions appeared shortly after the
introduction of the immunosuppressive drugs and
disappeared after these drugs were withdrawn. Similar
observations have been made regarding organ trans-
plantations,20 as well as nontransplant immunocom-
promised situations, such as recurrent leukemia,23
congenital immune deficiency,24 and immunosup-
pressive treatment for rheumatoid arthritis25 or bul-
lous pemphigoid.26 However, all these reports were
either descriptional or were based on circumstantial
evidence. To the best of our knowledge, the current
study is the first to provide statistically based evidence
for an independent association between immunosup-
pression and a more aggressive course of KS.
Much effort has been made through the years to
understand the scientific basis for the role of immu-
nosuppression in the development of KS.27–30 Some of
the hypotheses suggested may also be relevant to the
clinical behavior of the disease. For example, a dete-
riorated immune surveillance27 or an accelerated am-
plification of the lymphotropic HHV8 in an immuno-
compromised host,28 –30 or both, may represent the
essence of the pathogenesis of the disease and may be
relevant to the poorer outcome of these patients.
Age younger than 50 years was a favorable prog-
nostic factor, at least with regard to the risk of any
progression, in our study. We identified a relatively
small group of patients, 7.3% of our study population,
whose illness is characterized by a more indolent clin-
ical course than that of the general population of CKS
TABLE 4
Occurrence of any Progression: Multivariate Analysisa
Variables P value Risk ratio 95% CI
Age (continuous) 0.001 1.03 1.01–1.05
Gender 0.3 — —
Origin 0.3 — —
Second primary 0.6 — —
Immunosuppression 0.01 2.42 1.2–1.48
Anatomic distribution 0.06 — —
No. of lesions 0.6 — —
Observation 0.5 — —
CI: confidence interval.
a Progression includes any local or diffuse spread of lesions, as well as evidence of any development
requiring initiation or change in treatment. Multivariate analysis: Number of available patients � 213;
number of events � 94; overall score � 14.07; P � 0.0009.
TABLE 5
Dissemination of CKS Lesions: Multivariate Analysisa
Variables P value Risk ratio 95% CI
Age (continuous) 0.9 — —
Gender 1.0 — —
Origin 0.9 — —
Second primary 0.8 — —
Immunosuppression 0.006 4.6 1.5–13.7
Anatomical distribution 0.8 — —
Number of lesions 0.5 — —
Observation 0.9 — —
CKS: classical Kaposi sarcoma; CI: confidence interval.
a Dissemination included spread of CKS lesions from limited to diffuse disease. Multivariate analysis:
number of available patients � 130; number of events � 23; overall score � 8.94; P � 0.0028.
1986 CANCER November 1, 2002 / Volume 95 / Number 9
patients. Although young age is recognized as a good
prognostic factor among patients with a variety of
solid31 and hematologic malignancies,32 we are not
aware of any previous reports that suggest such an
influence among CKS patients. We believe that devel-
opment of CKS, a disease that is strongly associated
with older patients, in young patients who are not
necessarily immunocompromised might represent a
distinct clinical entity. The clinical course can be just
one of the features that distinguish between the two.
Another prominent finding of the current study is
that in none of the various univariate models tested
was observation alone (applied in 31% of our patients)
an unfavorable factor for the outcome of CKS. This
may confer more confidence to the physician who
chooses the approach of “watchful waiting” as pri-
mary management of selected patients with CKS.
In conclusion, we analyzed a large group of CKS
patients with clinical features resembling those de-
scribed by others and found that immunosuppression
and an age of 50 years or older were strongly associ-
ated with poorer outcome of the disease. We also
suggest that the two end points used in the current
study may serve for future prognostic factor analysis.
REFERENCES
1. Wahman A, Melnick SL, Rhame FS. The epidemiology of
classic, African and immunosuppressed Kaposi’s sarcoma.
Epidemiol Rev. 1991;13:178 –199.
2. Friedman-Kein A, Saltzman BR. Clinical manifestations of
classical endemic African and epidemic AIDS-associated
Kaposi’s sarcoma. J Am Acad Dermatol. 1990;22:1237–1250.
3. Stein ME, Kantor A, Spencer D, Bezwoda WR. Classical
Kaposi’s sarcoma in Caucasians in Africa — experience at
the Johannesburg Hospital (1978-1992). Dermatology. 1994;
188:182–187.
4. Sarid R, Olsen SJ, Moore PS. Kaposi’s sarcoma-associated
herpes virus: epidemiology, virology and molecular biology.
Adv Virus Res. 1999;52:139 –232.
5. Friedman-Birnbaum R, Weltfriend S, Kats I. Kaposi’s sar-
coma: retrospective study of 67 cases with the classical form.
Dermatologica. 1990;180:13–17.
6. Punesmith TH, Shrum JP. Kaposi’s sarcoma. Int J Dermatol.
1994;33:755–762.
7. Brenner B, Rakowsky E, Katz A, et al. Tailoring treatment for
classical Kaposi’s sarcoma: comprehensive clinical guide-
lines. Int J Oncol. 1999;14:1097–1102.
8. Kaplan EL, Meier P. Non-parametric estimation from in-
complete observation. J Am Stat. 1958;53:457– 481.
9. Cox DR. Regression models and life tables. J R Stat Soc [B].
1972;34:187–220.
10. Safai B, Mike V, Giraldo G. Association of Kaposi’s sarcoma
with second primary malignancies. Cancer. 1980;45:1472–
1479.
11. Feuerman EJ, Potruch-Eisenkraft S. Kaposi’s sarcoma. A fol-
low-up of 38 patients. Dermatologica. 1973;146:115–122.
12. Weissmann A, Linn S, Weltfriend S, Friedman-Birnbaum R.
Epidemiological study of classic Kaposi’s sarcoma: a retro-
spective review of 125 cases from Northern Israel. J Eur Acad
Dermatol Venereol. 2000;14:91–95.
13. Fenig E, Brenner B, Rakovsky E, Lapidoth M, Katz A, Sulkes
A. Classic Kaposi sarcoma. Am J Clin Oncol. 1998;21:498 –
500.
14. Iscovich J, Boffetta P, Winkelmann R, Brennan P. Classic
Kaposi’s sarcoma as a second primary neoplasm. Int J Can-
cer. 1999;80:178 –182.
15. Fossati S, Boneschi V, Ferrucci S, Brambilla L. Human im-
munodeficiency virus-negative Kaposi’s sarcoma and lym-
phoproliferative disorders. Cancer. 1999;85:1611–1615.
16. Piette WW. The incidence of second malignancies in subsets
of Kaposi’s sarcoma. J Am Acad Dermatol. 1987;16:855– 861.
17. Klepp O, Dahl O, Stenwig JT. Association of Kaposi’s sar-
coma and prior immunosuppressive therapy. Cancer. 1978;
42:2626 –2630.
18. Penn I. Kaposi’s sarcoma in organ transplant recipients.
Transplantation. 1979;17:8 –11.
19. Krown SE, Metroka C, Wernz JC. Kaposi’s sarcoma in the
acquired immunodeficiency syndrome: a proposal for uni-
form evaluation, response and staging system criteria. J Clin
Oncol. 1989;7:1201–1207.
20. Yokois NU, Perlman EJ, Colombina P, Wise B, Schwarz KB.
Kaposi’s sarcoma presenting as a protracted multisystem
illness in an adolescent liver transplantrecipient. Liver
Transplant Surg. 1997;3:541–544.
21. Toth A, Alfody F, Jaray J, et al. Disseminated Kaposi’s sar-
coma in immunosuppressed patients. Acta Chir Hung. 1995;
35:53– 62.
22. Haim S, Shafrir A, Better OS, Robinson E, Chaimovitz C,
Erlik D. Kaposi’s sarcoma in association with immunosup-
pressive therapy. Isr J Med Sci. 1972;8:1993–1997.
23. Roszkiewicz A, Roszkiewicz J, Lange M, Tukaj C. Kaposi’s
sarcoma following long-term immunosuppressive therapy:
clinical, histologic, and ultrastructural study. Cutis. 1998;61:
137–141.
24. Kusenbach G, Rubben A, Schneider EM, et al. Herpes virus
(KSHV) associated Kaposi sarcoma in a 3-year-old child with
non-HIV-induced immunodeficiency. Eur J Pediatr. 1997;
156:440 – 443.
25. Casoli P, Tumiati B. Kaposi’s sarcoma, rheumatoid arthritis
and immunosuppressive and/or corticosteroid therapy.
J Rheumatol. 1992;19:1316 –1317.
26. Halpern SM, Parslew R, Cerior R, Kirby JT, Sharpe GR. Ka-
posi’s sarcoma associated with immunosuppression for bul-
lous pemphigoid. Br J Dermatol. 1997;137:140 –143.
27. Doll R, Kinlen I. Immunosurveillance and cancer: epidemi-
ological evidence. Br Med J. 1970;4:420 – 422.
28. Doutrelepont JM, De Pauw L, Gruber SA, et al. Renal trans-
plantation exposes patients with previous Kaposi’s sarcoma
to a high risk of recurrence. Transplantation. 1996;62:463–
466.
29. Siegel B, Levinton-Kriss S, Schiffer A, et al. Kaposi’s sarcoma
in immunosuppression. Possibly the result of a dual viral
infection. Cancer. 1990;65:492– 498.
30. Moore PS, Chang Y. Detection of herpes virus-like DNA
sequences in Kaposi’s sarcoma in patients with and without
HIV infection. N Engl J Med. 1995;332:1181–1185.
31. Markman M, Lewis JL, Saigo P, et al. Impact of age on
survival of patients with ovarian cancer. Gynecol Oncol.
1993;49:236 –239.
32. Shipp MA, Harrington DP, Anderson JR, et al. A predictive
model for aggressive non-Hodgkin’s lymphoma. N Engl
J Med. 1993;329:987–994.
Classical Kaposi Sarcoma/Brenner et al. 1987