Cancer-associated thrombosis (CAT) - New challenges for the hematologist - Section 2

Clinical prediction scores for venous thromboembolism in patients

with a hematological malignancy
Frits I. Mulder,1,2 Nick van Es1
1
Department of Vascular Medicine, Academic Medical Center, Amsterdam; 2Department of Internal Medicine,
Tergooi hospital, Hilversum, The Netherlands

Take Home Messages

- The performance of the Khorana score in predicting venous thromboembolism in patients with a hematological malig-
nancy is uncertain.
- ThroLy is a promising score for prediction of arterial or venous thromboembolism in lymphoma patients, but needs fur-
ther external validation.
- The clinical benefit of all risk scores remains uncertain, since they have not been thoroughly evaluated in randomized
controlled trials of risk-adapted thromboprophylaxis.

Introduction was derived in a prospective cohort of 2,701 cancer patients, of

Venous thromboembolism (VTE) is a common complication in
patients with hematological malignancy with an overall inci-
dence of approximately 3.5% per 100 person years1. The risk of
VTE varies substantially across different types of hematological
malignancy, and also depends on cancer treatment, presence of
a central venous catheter, and cancer stage.2 International guide-
lines recommend routine thromboprophylaxis in patients with
multiple myeloma treated with immunomodulatory chemother-
apy,3,4 but not in those with other hematological malignancies.
Several risk scores have been developed to identify cancer
patients at high risk of VTE in whom thromboprophylaxis may
be justified, but only a few were specifically evaluated in patients
with a hematological malignancy. This review aims to provide
an overview of the performance and potential clinical utility of
the currently available risk scores for cancer-associated VTE in
patients with a hematological malignancy.
Khorana score
The most widely used risk assessment tool for cancer-associated
VTE is the Khorana score, which is now endorsed by several
guidelines.5,6 This risk score aims to identify ambulatory patients
with solid cancer or lymphoma at high risk of VTE prior to can-
cer therapy based on five clinical and laboratory parameters:
primary tumor site (e.g. +1 point for lymphoma), platelet count
of 350×109/L or more (+1 point), hemoglobin concentration of
less than 10 g/dL and/or use of erythropoiesis-stimulating agents
(+1 point), leukocyte count of more than 11×109/L (+1 point),
and body mass index of 35 kg/m2 or more (+1 point).7 The score
whom 12% patients had lymphoma. In the internal validation
cohort of 1,365 patients, the observed 3-month VTE risk was
6.7% in patients with a high-risk score (≥3 points) compared to
1.5% in patients with a low or intermediate risk score (0-1
points), corresponding to a relative risk (RR) of 4.5 (95% con-
fidence interval [CI], 2.1-9.6; Table 1). Results were not report-
ed separately for lymphoma patients, which limits the generaliz-
ability of the findings for this particular group. In addition, the
score was not intended to be used in patients with other types of
hematological malignancies. In the past 3 years, several studies
addressed these issues by evaluating the performance of the
Khorana score in patients with various hematological malignan-
cies, as summarized in Table 1. Taken together, these external
validation studies show that the relative risk of VTE in patients
with a hematological malignancy and a high-risk Khorana score
is lower (RR 1.2 to 2.4) than reported in the original derivation
study. Whether this is due to differences in case-mix, study
design, or follow-up duration, or reflects a lower performance in
patients with a hematological malignancy in general remains
unknown. Another concern is the low proportion of patients
classified as high risk, ranging between 5 and 15%, which, in
combination with the modest relative risks, limits the clinical
utility of the score in selecting patients for thromboprophylaxis.
The reported positive predictive values (7 to 19%) may be high
enough to justify thromboprophylaxis, but differences in follow-
up duration and types of included cancers make it difficult to
translate these findings to clinical practice (Table 1). The score
has not been validated in patients with acute leukemia, because
these patients often have abnormal blood counts. Whether
extending the Khorana score with D-dimer and soluble

Educational Updates in Hematology Book | 2018; 2(S2) | 37 |

F.I. Mulder, N. van Es Clinical prediction scores for venous thromboembolism in patients with a hematological malignancy

P-selectin levels, as proposed by Ay and colleagues, improves
risk stratification in lymphoma patients is unknown.8
ThroLy score
The ThroLy score was developed by Antic and colleagues for
predicting venous and arterial thrombotic events in ambulatory
lymphoma patients.9 The score is composed of seven clinical and
laboratory variables and was derived in 1,236 patients with lym-
phoma or chronic lymphocytic leukemia receiving chemothera-
py. In the internal validation cohort of 584 patients, the risk of
arterial or venous thromboembolism was 29% in patients with
an intermediate or high-risk score (≥2 points) compared to 2.4%
in those with a low risk score (0-1 points). The c-statistic of the
score was 0.86. However, important information needed to fully
appreciate the score’s performance, such as follow up-duration
and proportion of patients classified as being at risk, were unfor-
tunately not reported. Both venous and arterial thromboembolic
events, including superficial thrombophlebitis, were part of the
outcome, which limits the use of the score when deciding about
primary VTE prevention. Although the ThroLy score appears
promising, it needs to be validated in external patient cohorts
before it can be implemented in clinical practice.
IPSET-thrombosis score
The International Prognostic Score for Essential
Thrombocythemia (IPSET) score was developed to predict arte-
rial or venous thrombotic events in patients with essential
thrombocythemia (ET).10 The score is based on four variables:
age 60 years or older (+1 point), cardiovascular risk factors (+1
point), previous thrombosis (+2 points), and JAK2V617F muta-
tion (+2 points). The risk of arterial or venous thrombotic event
in the high-risk group (≥3 points) ranged between 13 and 23%
in the original derivation, internal validation, and external vali-
dation cohorts. Several external validation studies confirmed the
discriminatory performance of the IPSET score with reported
relative risks of thrombosis between 3.0 and 15 in high risk
patients (Table 1). The group that introduced the IPSET-score
subsequently proposed a ‘practice-relevant revision’ by stratify-
ing patients into four groups based on age, previous thrombosis,
and JAK2V617F mutation.11 Although the score appeared to
perform well in an external validation study,12 its ability to
improve risk-adapted prophylaxis needs to be confirmed in
prospective studies.
Thromboprohylaxis risk assessment tool for multiple
myeloma
A VTE risk assessment tool was developed by the International
Multiple myeloma Working Group (IMWG) for patients with
multiple myeloma receiving immunomodulatory drugs.4 The tool
consists of a list of 17 patient-, myeloma-, and treatment-specific
risk factors. If no or any one of the risk factors are present in a
patient receiving thalidomide or lenalidomide, aspirin once daily
is recommended. If two or more risk factors are present, a pro-
phylactic dose of low-molecular-weight heparin or warfarin tar-
geted at an INR of 2 to 3 is recommended. The IMWG risk score
is based on expert opinion without any validation studies, and
should therefore not be regarded a firm guideline.
Future perspectives
The Khorana score has now been evaluated in several studies of
patients with hematological malignancies, but results are not
unambiguous. Given the uncertainty, clinicians should be cau-

Table 1. Studies evaluating risk scores for thrombosis in patients with a hematological malignancy.

First author Number of Type Follow-up Outcome Overall Patients in high Positive RR for high vs
patients duration incidence risk group predictive value lower risk patients
n (%) n (%) (%) (95% CI)

Khorana score
Khorana 2008 (derivation cohort)7 2701* Lymphoma Median 73 days VTE 60 (2.2)* 340 (12.6)* 7.1* 4.6 (2.8-7.7)*
Khorana 2008 (validation cohort) 1365† Lymphoma Median 73 days VTE 28 (2.1)† 149 (10.9)† 6.7† 4.5 (2.1-9.6)†
Lim 201514 322 DLBCL Median 42 months VTE 34 (10.6) 16 (4.9) 18.8 1.9 (0.6-5.4)
Antic 2016 (cohort 1)15 1236 NHL, HL, CLL, SLL NR ATE, VTE 65 (5.3) NR 14.8 NR
Antic 2016 (cohort 2)15 584 NHL, HL, CLL, SLL NR ATE, VTE 34 (5.8) NR 14.8 NR
Rupa-Matysek 201716 428 DLBCL, HL Median 37 months VTE 64 (15) 64 (15) 17 1.2 (0.7-2.1)
Santi 201717 1189 DLBCL, FL, INFL, MCL 6 months VTE 41 (3.4) 141 (11.9) 7.1 2.4 (1.2-4.8)
ThroLy score
Antic 20169 (derivation cohort) 1236 NHL, HL, CLL, SLL NR ATE, VTE 65 (5.3) NR 25.1 NR
Antic 2016 (validation cohort) 584 NHL, HL, CLL, SLL NR ATE, VTE 34 (5.8) NR 28.9 NR
IPSET thrombosis score
Barbui 2012 (derivation cohort)18 535 ET Median 6.5 years ATE, VTE 59 (11) 63 (12) 17.5 1.7 (0.9-3.1)
Barbui 2012 (internal validation cohort)18 356 ET Median 6.1 years ATE, VTE 44 (12.4) 48 (14) 22.9 2.1 (1.2-3.9)
Barbui 2012 (external validation cohort)18 329 ET Median 5.0 years ATE, VTE 31 (9.4) 164 (50) 12.8 2.1 (1.0-4.4)
Fu 201419 746 ET Median 49 months ATE, VTE 77 (10.3) 221 (29.6) 19.5 3.0 (2.0-4.6)
Sevindik 201520 112 ET Median 4.1 years ATE, VTE 38 (33.9) 45 (40.2) 62.2 4.2 (2.3-7.7)
Navarro 201621 44 ET NR ATE, VTE 19 (43.2) 24 (54.6) 25.0 15.0 (2.2-102.8)
*Of the total study group, 12% had lymphoma and 88% a solid cancer. † Of the total study group, 14% had lymphoma and 86% a solid cancer. Abbreviations: ATE, arterial thromboembolism; CLL, chronic lym-
phocytic lymphoma; DLBCL, diffuse large B-cell lymphoma; ET, essential thrombocythemia; FL, follicular lymphoma; HL, Hodgkin’s lymphoma; INFL, indolent non-follicular lymphoma; NHL, non-Hodgkin lymphoma;
MCL, mantle cell lymphoma; MM, multiple myeloma; NR; not reported; RR; relative risk; SLL, small lymphocytic lymphoma; VTE, venous thromboembolism.

| 38 | Educational Updates in Hematology Book | 2018; 2(S2)

F.I. Mulder, N. van Es Clinical prediction scores for venous thromboembolism in patients with a hematological malignancy
tious to use the score to select patients for thromboprophylaxis.
The lymphoma-specific ThroLy score and ET-specific (revised)
IPSET-scores are promising, but need additional external valida-
tion before they can be used in clinical practice. Future studies
need to focus either on validating existing scores in different set-
tings or on developing new models that more accurately select
patients with a high short-term risk of VTE. Ideally, scores
should then be used to guide risk-adapted thromboprophylaxis
in randomized trials to confirm their performance and clinical
benefit.
References
1. Horsted F, West J, Grainge MJ. Risk of venous thromboembolism
in patients with cancer: a systematic review and meta-analysis.
PLoS Med 2012;9:e1001275.
2. Caruso V, Di Castelnuovo A, Meschengieser S, et al. Thrombotic
complications in adult patients with lymphoma: A meta-analysis of
29 independent cohorts including 18 018 patients and 1149 events.
Blood 2010;115:5322-8.
3. Terpos E, Kleber M, Engelhardt M, et al. European myeloma net-
work guidelines for the management of multiple myeloma-related
complications. Haematologica 2015;100:1254-66.
*4. International Myeloma Working Group (IMWG). IMWG
Guidelines for the Prevention of Thalidomide- and Lenalidomide-
Associated Thrombosis in Myeloma 2010.
Expert-opinion based list of risk factors for venous thromboem-
bolism in patients with multiple myeloma receiving immunomodu-
latory drugs that is used to guide decisions about primary preven-
tion with aspirin, low-molecular-weight heparin, or vitamin K
antagonists.
5. Lyman GH, Bohlke K, Khorana AA, et al. Venous thromboem-
bolism prophylaxis and treatment in patients with cancer:
American society of clinical oncology clinical practice guideline
update 2014. J Clin Oncol 2015;33:654-56.
6. Streiff MB, Holmstrom B, Ashrani A, et al. Cancer-associated
venous thromboembolic disease, version 1.2015: Featured updates
to the NCCN Guidelines. JNCCN J Natl Compr Cancer Netw
2015;13:1079-95.
7. Khorana AA, Kuderer NM, Culakova E, , et al Development and
validation of a predictive model for chemotherapy-associated
thrombosis. Clin Trials Obs 2008;111:4902-8.
8. Ay C, Dunkler D, Marosi C, et al. Prediction of venous thromboem-
bolism in cancer patients. Blood 2010;116:5377-82.
*9. Antic D, Milic N, Nikolovski S, et al. Development and validation
of multivariable predictive model for thromboembolic events in
lymphoma patients. Am J Hematol 2016;91:1014-1019.
Derivation and internal validation of a novel risk score (ThroLy)
for arterial and venous thromboembolism in patients with lym-
phoma, which requires external validation before it should be used
in clinical practice.
*10. Barbui T, Finazzi G, Carobbio A, et al. Development and validation
of an International Prognostic Score of thrombosis in World Health
Organization-essential thrombocythemia (IPSET-thrombosis).
Blood 2012;120:5128-33.
Derivation, internal validation, and external validation of a clinical
risk score (IPSET-thrombosis) for arterial and venous thromboem-
bolism that can be used to guide decisions about providing aspirin
in patients with essential thrombocythemia.
*11. Barbui T, Vannucchi AM, Buxhofer-Ausch V, et al. Practice-relevant
revision of IPSET-thrombosis based on 1019 patients with WHO-
defined essential thrombocythemia. Blood Cancer J 2015;5:2-4.
A revision of the IPSET-score that aims to further improve risk
stratification and simplify clinical decision making about aspirin in
patients with essential thrombocytemia.
12. Haider M, Gangat N, Lasho T, et al. Validation of the revised inter-
national prognostic score of thrombosis for essential thrombo-
cythemia (IPSET-thrombosis) in 585 Mayo clinic patients. Am J
Hematol 2016;91:390-4.
*13. Khorana AA, Kuderer NM, Culakova E, et al. Development and
validation of a predictive model for chemotherapy-associated
thrombosis. Blood 2008;111:4902-7.
Derivation and internal validation of a well-known clinical risk
score (Khorana score) for venous thromboembolism in ambulatory
patients with solid cancer or lymphoma that may guide decisions
about thromboprophylaxis.
14. Lim SH, Woo SY, Kim S, et al. Cross-sectional Study of patients
with diffuse large B-cell lymphoma: Assessing the effect of host sta-
tus, tumor burden and inflammatory activity on venous throm-
boembolism. Cancer Res Treat 2015:1-10.
15. Antic D, Milic N, Nikolovski S, et al. Comparative analysis of pre-
dictive models for thromboembolic events in lymphoma patients.
Hematol Oncol 2017;35:416.
16. Rupa-Matysek J, Gil L, Ka mierczak M, et al. Prediction of venous
thromboembolism in newly diagnosed patients treated for lym-
phoid malignancies: validation of the Khorana Risk Score. Med
Oncol 2018;35:1-8.
17. Santi RM, Ceccarelli M, Bernoccol E, et al. Khorana score and his-
totype predicts incidence of early venous thromboembolism in non-
Hodgkin lymphomas. Thromb Haemost 2017;117:1615-21.
18. Barbui T, Finazzi G, Carobbio A, et al. Development and validation
of an International Prognostic Score of thrombosis in World Health
Organization – essential thrombocythemia (IPSET-thrombosis).
Blood 2015;120:5128-34.
19. Fu R, Xuan M, Lv C, et al. External validation and clinical evalua-
tion of the International Prognostic Score of Thrombosis for
Essential Thrombocythemia (IPSET-thrombosis) in a large cohort of
Chinese patients. Eur J Haematol 2014;92:502-9.
20. Sevindik OG, Mersin S, Katgi A, et al. IPSET-thrombosis better
identifies thrombosis-free survival: A Turkish cohort. Clin
Lymphoma, Myeloma Leuk 2015;15:e101-4.
21. Navarro LM, Trufelli DC, Bonito DR, et al. Application of prog-
nostic score IPSET-thrombosis in patients with essential thrombo-
cythemia of a Brazilian public service. Rev Assoc Med Bras
2016;62:647-51.
Educational Updates in Hematology Book | 2018; 2(S2) | 39 |