Clinical factors associated with venous thromboembolism in multiple myeloma patients treated with daratumumab

Abstract

Background

Venous thromboembolism (VTE) is a common cause of morbidity and mortality in patients with multiple myeloma (MM). There has been little evidence investigating the factors associated with VTE during therapy with daratumumab-containing regimens.

Objectives

Our study examined the association of different clinical, medication- and procedure-related factors on the development of VTE events.

Methods

We performed a retrospective chart review of multiple myeloma patients treated at our institution with daratumumab-containing regimens and who developed VTE within 12 months of starting daratumumab. We collected the information on clinical factors, cancer therapy, antithrombotic medications, and central venous access procedures during the observation period. A univariate logistic regression was done first to identify variables associated with VTE, after which those variables were entered into a multivariate analysis.

Results

Among patients with MM who received a daratumumab-containing regimen, the incidence of VTE was 9.6% at 12 months from the initiation of daratumumab. We found that the number of PICC line insertions and the presence of multiple medical comorbidities were associated with VTE development.

Conclusions

We conclude that VTE is frequent during therapy for MM, particularly in the presence of certain risk factors such as use of multiple PICC line insertions and recent infections. Our observations do not imply causality and limited by the retrospective design, small number of VTE events and lack of time-to-event analysis in the study.

Introduction

Hematologic malignancies are significantly associated with risk of venous thromboembolism (VTE), in particular, multiple myeloma (MM). One study showed that patients with MM had a 7.5-fold increase in development of VTE within 1 year of diagnosis compared to their matched controls [18]. This risk is particularly enhanced within the first six months after diagnosis, with one study noting that multiple myeloma had a 20.3-fold increase in development of VTE during this timeframe compared to their matched controls [22]. It is understood that VTE risk is associated with specific cancer therapy regimens for MM, such as high doses of steroids, chemotherapy, and immunomodulatory agents (iMiDs). The use of therapies such as thalidomide in combination with dexamethasone produces a much greater risk for VTE formation. One randomized trial found that patients with MM given thalidomide/dexamethasone compared to placebo/dexamethasone showed a significant difference (17% to 3%, respectively) in VTE formation [25]. Other treatments for MM include newer generations of thalidomide derivatives (lenalidomide, pomalidomide), which have been shown to increase the incidence of VTE. The combination of either iMiD with dexamethasone increased the incidence of VTE to above 15–20% compared to the normal incidence of 3–4% [24]. Collectively, those data and evidence from other studies have been used to create risk assessment models for VTE, review and consensus statements for thromboprophylaxis in MM [1, 9, 11, 12, 16, 29]. For example, the IMPEDE score identified that for multiple myeloma patients on iMiDs; Asian/Pacific Islander ethnicity, an elevated body mass index, the history of pelvic, hip-femoral fractures, the use of erythropoietin stimulating agents (ESA), the use of dexamethasone and/or doxorubicin, prior history of VTE, and the presence of tunneled -line central venous catheters (CVC) were associated to the risk of VTE [27]. Similarly, the SAVED score identified in iMiD treated MM patients similar overlapping features to the IMPEDE score, besides age > 80 years, as consistent risk factors associated to VTE [19]. Further VTE risk assessment models for MM patients treated with iMiD have incorporated tumor characteristics, such as cytogenetic abnormalities, into the contributing VTE risk factors [7]. Such VTE risk assessment model tools have been extensively validated and compared for performance [10, 12, 15, 17].

The current standard of care for MM has since evolved. Combination drug regimens that include thalidomide derivatives, proteasome inhibitors, steroids, and different monoclonal antibodies are the pillars of the management of myeloma.

This study aims to study clinical, tumor-related, treatment-related, and comorbidity factors associated with the development of VTE in MM patients on daratumumab-containing treatment regimens.

Methods

This was a retrospective review of medical records of patients who received daratumumab, registered in a data warehouse designed to study the interference of monoclonal antibodies and blood typing procedures, at MD Anderson Cancer Center (MDACC) between April 2014 and February 2021. The study was approved by the institutional review board at MDACC. A waiver for informed consent was provided as this retrospective study could not be practicably conducted without it, as the patients may be lost to follow-up, may no longer be undergoing treatment, or may have died. Eligible patients were adult individuals with multiple myeloma and treated with at least one dose of daratumumab. Included patients were followed up at MDACC with available clinical data within the following 12 months after the first dose of daratumumab. We excluded patients who were prescribed long-term anticoagulation at the initiation of daratumumab (e.g., atrial fibrillation). All medical records were screened for deep venous thrombosis (DVT) and pulmonary embolism (PE) through Doppler ultrasound and contrast-enhanced radiology reports, respectively, within 365 days after the date of the first dose of daratumumab. A natural language processing (NLP) algorithm was used to determine the likely VTE-positive cases, with each case confirmed with manual chart review. Patients were divided into two categories, based on the occurrence of VTE diagnosis after to the first dose of daratumumab: VTE post-daratumumab versus non-VTE post-daratumumab. The co-variates of interest were selected from three previous VTE risk stratification systems (SAVED, IMPEDE-VTE, and IMWG) [1, 9, 12]. Variables of interest included: history of surgery within 90 days prior to daratumumab, race, ethnicity, prior history of VTE, comorbidities, age, use of dexamethasone, use of immunomodulatory drug (iMiD), history of pathologic fracture, use of erythropoietin stimulating agent (ESA), use of doxorubicin, presence of venous catheter, use of antithrombotics and anticoagulants, and the occurrence of active infection. We also analyzed the individual effect of each variable in one of the combined IMWG variables corresponding to comorbidities (cardiac disease, chronic kidney disease, acute infection, and immobilization) on the development of VTE; the presence of each medical comorbidity was assessed via manual chart review. Cardiac disease was defined as one of the following: hypertension, congestive heart failure, atrial fibrillation, or coronary artery disease. Immobilization was defined as either a prolonged hospitalization or a stay at a skilled nursing facility (SNF). Additionally, we captured data on myeloma disease staging, prior history of autologous stem cell transplant, and disease status at the time of daratumumab initiation.

All variables of interest were analyzed using descriptive statistics. Differences in all those variables were compared within the two category-VTE groups. Chi-square test or Fisher’s exact test was used to assess the association between categorical variables, and the Wilcoxon rank sum was used for continuous numeric variables. We decided to focus on the factors associated with the development of VTE, instead of a time-to-event analysis, which would require data that were not available in a retrospective collection of information. Univariate logistic regression was used to assess the association between the development of VTE and demographic and clinical factors of interest. Subsequently, a multivariable logistic regression analysis was performed, which considered factors that were significant in univariate analysis.

All statistical analyses were performed using R version 4.2.2. All statistical tests used a significance level of 5%. No adjustments for multiple testing were made.

Results

A total of 204 patients who received daratumumab were initially screened for inclusion in the study. 15 of these patients did not have a diagnosis of multiple myeloma, and one patient had a diagnosis of primary amyloidosis; those cases were excluded from the analyses. The median time to follow-up for our patient cohort was 1.3 years. A total of 82 patients (43.6%) died within the 12-month observation period of our study.

The final cohort of patients (n = 188) with multiple myeloma and treated with daratumumab were middle-aged to elderly (median age, 65 years), and 54.3% of patients were male. The population was predominantly white (65%) of non-Hispanic origin (81%). Almost all patients had relapsed/refractory malignant disease (94%) with a median of 5 prior lines of therapy (LOT) before the first dose of daratumumab. Dexamethasone was used in 98.4% of the patients. Other malignant diseases and details of prior LOT are detailed in Table 1.

Table 1.

Baseline characteristics of patient population

Patient Characteristic	N (%)
Age (years), median (IQR)	65 (58–72)
Gender
Female	86 (45.7)
Male	102 (54.3)
Ethnicity
Hispanic/Latino	27 (14.4)
Non-Hispanic/Latino	152 (81.3)
Unknown	8 (4.3)
Race
Asian	7 (3.7)
Black/African American	36 (19.1)
Declined to Answer	1 (0.5)
Other	18 (9.6)
Unknown	3 (1.6)
White/Caucasian	123 (65.4)
ECOG performance status
0	38 (20.2)
1	106 (56.4)
2	29 (15.4)
3	13 (6.9)
4	2 (1.1)
ISS stage
I	55 (34.6)
II	69 (43.4)
III	35 (22.0)
Prior ASCT	130 (69.2)
Multiple Myeloma Disease Status
Treatment-naïve	6 (3.2)
Stable/remission	5 (2.6)
Relapsed/refractory	175 (94.1)
Number of prior LOT, median (IQR)	5 (3–8)
Prior lytic lesion	110 (58.5)
Prior pathologic fracture	91 (48.4)
High risk features
High risk cytogenetics*	91 (48.4)
Other high-risk features†	158 (84.0)
M-protein at start of daratumumab (g/dL), median (IQR)	0.50 (0.10–1.50)
History of VTE prior to daratumumab	40 (21.3)
Time from cancer diagnosis to start of daratumumab (years), median (IQR)	4.0 (1.7–7.2)
Use of dexamethasone	185 (98.4)
ESA use	4 (2.1)
Use of doxorubicin	23 (12.2)
Use of alkylating chemotherapy	51 (27.1)
Use of PI	30 (16.0)
Use of iMiD	50 (26.6)
Antiplatelet use	95 (50.5)
Any anticoagulant use	148 (78.7)
Therapeutic anticoagulant use	10 (5.3)
Venous catheter placement
Any CVC or PICC placement	96 (51.1)
CVC placement	53 (28.2)
PICC placement	49 (26.1)
Number for PICC insertions
0	139 (73.9)
1	37 (19.6)
2	10 (5.3)
3	2 (1.1)
Pacemaker placement prior to daratumumab	6 (3.2)
Obesity	14 (7.5)
BMI, median (IQR)	27.6 (23.8–28.3)
Cardiac disease	31 (16.5)
CKD	21 (11.2)
Acute infection	13 (6.9)
Immobilization	4 (2.1)
Other thrombophilia	0

^*High-risk cytogenetics were defined as one of the following: t(4;14), t(14,16), del(17p), gain(1q21), monosomy 17, monosomy 13, complex karyotype, or tetrasomy

^†Other high-risk features were defined as one of the following: extramedullary disease, circulating plasma cells, renal failure, thrombocytopenia (platelet count < 150,000/µL), free light chains > 100 mg/L, or elevated lactate dehydrogenase (LDH)

IQR interquartile range, ECOG Eastern Cooperative Oncology Group, iMiD immunomodulatory agent, Pl proteosome inhibitor, ASCT autologous stem cell transplant, LOT lines of therapy, VTE venous thromboembolism, ESA erythropoietin stimulating agent, CVC central venous catheter, PICC peripherally inserted central catheter, CKD chronic kidney disease, RBC red blood cell, ISS International Staging System

In regard to risk factors for VTE, the use of central venous catheters (CVC) was present in 51% of the population, with half of the CVCs corresponding to peripherally inserted devices. Chronic cardiac and renal comorbidities were present in 16.5% and 11.2% of the population, respectively. The concurrent use of supportive erythropoietin-stimulating agents occurred in 2% of patients. The population did not have any other thrombophilic disorders documented in their personal history data.

Acute venous thromboembolism occurred in 18 patients (9.6%) within the following 12 months from the first daratumumab dose administration. The individual patient-VTE events included: nine as lower extremity DVT/PE, eight as upper extremity DVT (50% of those events were associated with central venous catheters), and one located in the inferior vena cava. The median time to VTE from the first dose of daratumumab was 3.9 months (IQR 1.6–6.1 months).

The initial univariate analysis showed a similar distribution of demographic, cancer-stage, and cancer-therapy characteristics within the cohort. Amongst co-variates from the SAVED, IMPEDE-VTE, and IMWG scoring systems, those associated with VTE were the presence of comorbidities (acute infection, immobility; p = 0.0031 and 0.0357, respectively) and the number of peripherally inserted central line (PICC) insertion procedures (p = 0.0101) during the observation period (Tables 2 and 3).

Table 2.

Univariate logistic regression of various categorical variables associated with time to VTE based on published risk assessment models (n = 188)

Variable	VTE after daratumumab (n = 18)	No VTE post-daratumumab (n = 170)	Odds Ratio (95% CI)	P-value
Male sex	9	93	0.83 (0.31, 2.19)	0.7036
Hispanic/Latino ethnicity	3	24	Reference group	0.8180
Non-Hispanic/Latino ethnicity	15	137	0.79 (0.23, 2.72)
Unknown	0	8	0.41 (0.02, 8.82)
Asian	1	6	Reference group	0.9614
Black/African American	3	33	0.45 (0.05, 3.66)
Declined to Answer	0	1	1.44 (0.04, 56.14)
Other	2	16	0.66 (0.07, 6.04)
Unknown	0	3	0.62 (0.02, 19.59)
White/Caucasian	12	111	0.49 (0.07, 3.15)
ECOG 0	2	36	Reference group	0.5193
1	13	93	2.11 (0.52, 8.57)
2	1	28	0.77 (0.10, 6.17)
3	2	11	3.17 (0.49, 20.68)
4	0	2	2.92 (0.11, 78.83)
History of VTE	3	37	0.72 (0.20, 2.62)	0.6060
Dexamethasone use	18	167	0.77 (0.04, 15.6)	0.8706
ESA use	0	4	1.00 (0.05, 19.30)	1.0000
Doxorubicin use	0	21	0.19 (0.01, 3.23)	0.1340
Other cancer therapy use*	6	75	0.68 (0.23, 1.81)	0.4439
iMiD use	4	46	0.77 (0.24, 2.46)	0.6533
Antiplatelet use	7	88	0.59 (0.22, 1.60)	0.2972
Any anticoagulant use	12	136	0.50 (0.18, 1.43)	0.2108
PICC placement	4	45	0.79 (0.25, 2.54)	0.6917
CVC placement	4	49	0.71 (0.22, 2.25)	0.5453
Any CVC or PICC	8	88	0.75 (0.28, 1.98)	0.5545
Surgery within 90 days	3	10	3.20 (0.79, 12.90)	0.1333
Pacemaker	2	4	5.19 (0.88, 30.60)	0.1014
Obesity†	1	13	0.70 (0.09, 5.70)	0.7285
Medical comorbidities ††	13	35	10.00 (3.35, 30.00)	< 0.0001
Cardiac disease	6	25	2.90 (1.00, 8.44)	0.0635
CKD	4	17	2.57 (0.76, 8.70)	0.1548
Acute infection	5	8	7.79 (2.23, 27.2)	0.0031
Immobilization	2	2	10.50 (1.38, 79.6)	0.0357
ISS Stage I	6	49	Reference group	0.9782
II	7	62	0.92 (0.29, 2.92)
III	4	31	1.05 (0.28, 4.04)
Relapsed/refractory disease status	16	159	Reference group	0.4370
Stable/remission disease status	1	4	4.14 (0.57, 30.00)
Treatment-naïve disease status	1	5	2.64 (0.40, 17.2)
Prior pathologic fractures	8	83	0.84 (0.31, 2.23)	0.7234
Prior lytic lesions	12	98	1.47 (0.53, 4.10)	0.4553
High-risk cytogenetics‡	9	82	1.07 (0.41, 2.84)	0.8867
Other high-risk features§	14	144	0.63 (0.19, 2.07)	0.4638
ASCT prior to daratumumab	11	119	0.67 (0.25, 1.84)	0.4459

^†Defined as a body mass index (BMI) of at least 30 kg/m²

* Includes alkylating agent and proteasome inhibitors

^‡High-risk cytogenetics were defined as one of the following: t(4;14), t(14,16), del(17p), gain(1q21), monosomy 17, monosomy 13, complex karyotype, or tetrasomy

^§Other high-risk features were defined as one of the following: extramedullary disease, circulating plasma cells, renal failure, thrombocytopenia (platelet count < 150,000/µL), free light chains > 100, or elevated lactate dehydrogenase (LDH)

^¶Values with no odds ratio were used as the comparison standard with other values (e.g. relapsed/refractory was used as the comparison standard for the stable/remission and treatment-naïve disease statuses)

^††Medical comorbidities include the following: cardiac disease, chronic renal disease, acute infection, or immobilization

VTE venous thromboembolism, ESA erythropoietin-stimulating agent, iMiD immunomodulatory agent, PICC peripherally inserted central catheter, CVC central venous catheter, CKD chronic kidney disease, ISS International Staging System, PI proteosome inhibitor, ASCT autologous stem cell transplant

Table 3.

Univariate logistic regression of various continuous variables associated to VTE based on published risk assessment models (n = 188)

Variables in the Equation	Odds Ratio	Confidence Ratio		P value
		Lower	Upper
Age	1.01	0.97	1.06	0.6198
BMI	0.95	0.87	1.04	0.2374
Time from cancer diagnosis to start of daratumumab	0.91	0.79	1.05	0.1935
LOT prior to daratumumab	0.93	0.80	1.07	0.3057
M-protein at daratumumab	0.71	0.42	1.20	0.2024
Number of PICC insertions	2.12	1.20	3.75	0.0101
Number of CVC insertions	1.25	0.44	3.53	0.6748

BMI body mass index, LOT lines of therapy, ECOG Eastern Cooperative Oncology Group), PICC peripherally inserted central catheter, CVC central venous catheter

We explored correlations between the prescription/hospital use of antiplatelet agents and/or anticoagulants (primarily prophylactic doses) and the incidence of VTE and found no statistically significant differences (Table 2). We did not capture interruptions of prescribed antithrombotics or anticoagulants. Therefore, the effect of thromboprophylaxis compliance was not examined. There were 10 patients who were prescribed therapeutic anticoagulation during the study observation period, all after acute VTE during the study observation (Table 1).

In the multivariable logistic regression model analysis, the presence of comorbidities (OR 8.14, p = 0.0007) and several PICC insertion procedures (OR 2.056, p = 0.0276) retained their association with VTE. After accounting for these, having an acute infection added no significant predictive ability to the model (Table 4).

Table 4.

Multivariable logistic regression model for VTE in MM patients receiving daratumumab

Variables in the Equation	Odds Ratio	Confidence Ratio		P value
		Lower	Upper
Medical comorbidities*	8.140	2.402	27.589	0.0007
Acute infection	1.876	0.455	7.742	0.3884
Number of PICC insertions	2.056	1.083	3.903	0.0276

^*Comorbidities include the following: cardiac disease, chronic renal disease, acute infection or immobilization

CKD chronic kidney disease, PICC peripherally inserted central catheter

Discussion

Our results show a frequent incidence (9.6% at 12 months) of acute VTE after the first dose of daratumumab-containing regimens. It is important to note that our findings do not imply daratumumab as the direct causality, but rather an observation consistent with the VTE incidence published in studies of patients treated with multi-agent regimens [3, 5, 26]. While the VTE capture in our study was retrospective, we used several methods (manual chart review and NLP tools) to identify the events with the goal of reducing the number of missed or inappropriately captured events.

In our cohort, almost all patients were prescribed steroids at least once after the beginning of daratumumab; while more than half received other cancer therapy agents after the use of daratumumab, including doxorubicin, alkylating agents, immunomodulatory drugs, or proteasome inhibitors.

There have been several studies that have described the occurrence of VTE during daratumumab-containing regimens. In the POLLUX trial, the daratumumab arm was found to have a lower incidence of VTE (defined as DVT and PE) compared to the control arm (lenalidomide + dexamethasone alone) (7/283 events vs. 9/281 events, respectively, RR 0.78 [0.29–2.06]) [8]. In the CASTOR trial, the daratumumab arm was found to have a nearly equivalent VTE incidence compared with the control arm (bortezomib + dexamethasone alone) (2/240 events vs. 2/234 events, respectively, RR 0.98 [0.14–6.87]) [23]. In the ALCYONE trial, the daratumumab arm was associated with no VTE events, and thus a decreased incidence compared to the control arm (bortezomib + melphalan + prednisone alone) (0/350 events vs. 4/356 events, respectively, RR 0.11 [0.01–2.11]) [20]. In the MAIA trial, the daratumumab arm showed a higher incidence of VTE compared to the control arm (lenalidomide + dexamethasone alone) (16/364 events vs. 22/365 events, respectively, RR 0.74 [0.40–1.39]) [14]. In the CASSIOPEIA trial, the daratumumab arm had a statistically significantly lower incidence of VTE compared to the control arm (bortezomib + thalidomide + dexamethasone alone) (8/536 events vs. 20/538 events, respectively, RR 0.41 [0.18–0.92]) [21]. In the GRIFFIN trial, the daratumumab arm also showed a lower incidence of VTE compared to the control arm (lenalidomide + bortezomib + dexamethasone alone) (2/99 events vs. 5/102 events, respectively, RR 0.42 [0.08–2.14]) [30]. Another study by Calafiore et al. showed that in the ten multiple myeloma patients who were treated with daratumumab only, none of them developed VTE, indicating that by itself, daratumumab did not increase the incidence of VTE [4]. Some of these patients received different modalities of thromboprophylaxis (7 with aspirin, 2 with LMWH, 1 not treated).

Interestingly, the rate of VTE in daratumumab clinical trials (including the study comparator arms) is lower than expected for patients with MM, a population historically – and from other clinical trials – known to have a higher risk for venous thrombosis [6], which is multiplied by several fold when high-dose steroids and/or combination chemotherapy are used [5–7]. In our cohort, we found that the incidence of VTE after the use of daratumumab-containing regimens did not seem different when other cancer therapy agents were concurrently used. We understand this observation is a result of other factors contributing to the development of VTE. The great majority of VTE occurred within the first 6 months from the first dose of a daratumumab-containing regimen, with about half of the events located in the upper extremities. Those observations, in contrast with the incidence reported in the daratumumab clinical trials [13, 14, 20, 21, 23, 30], suggest that other factors in our cohort may have contributed to the VTE risk. We observed associations between the occurrence of VTE and existing comorbidities, the use of recurrent PICC line insertions, and the occurrence of acute infection. Data from other hematologic cancer populations have observed that PICC line for venous access carries a higher risk of VTE when compared to other type of CVCs [28]. Recent infection events have been identified as an independent risk factor for thromboembolism in large general population based-studies [8]. Other cancer, non-myeloma specific, cohorts have identified CVCs and recent infection episodes as risk factors for VTE[2].

Since daratumumab has become one of the most widely used first-line agents for multiple myeloma, it is important to recognize which patient-related factors are more closely associated with the development of VTE in these patients. Our study has shown that specific medical comorbidities and indwelling vascular access procedures have association with the development of VTE in daratumumab treated, relapsed/refractory MM patients.

There are limitations to our study. Given the retrospective nature of the study, we were not able to determine if initiating VTE prophylaxis in at-risk patients on daratumumab-based regimens could prospectively cause a decrease in the incidence of VTE. Another limitation of our study is that most of the patients in our cohort have disease relapsed/refractory to previous cancer treatments, and therefore, a heterogeneous group of patients in regard to MM treatments, previous lines of therapy, which could have biased our observations to a population inherently at a higher risk for VTE. Due to the design of our study and the multiple concurrent therapies that patients received (steroids, iMiDs, proteasome inhibitors, alkylating agents), we could not make inferences on the contribution of daratumumab to the VTE risk in our population.

Moreover, in our study we arbitrarily selected a 12 month-point for assessment of VTE. Due to high mortality in our cohort, longer follow-up to assess VTE events was not possible. Current randomized controlled study data on the occurrence of VTE in daratumumab treated patients have followed up patients between 12 and 48 months. To our knowledge, there are no other prospective or retrospective comparative data in this topic.

Finally, due to the modest size of our cohort and the number of acute VTE events, the risk estimates for the factors found associated with VTE may reflect imprecision and possible model instability that will require confirmation with future larger studies.

VTE prevention in MM has several unmet needs, which include risk stratification, standardized prophylaxis strategies, education, and awareness. Although risk assessment tools exist, they do not capture all aspects of the individual patient’s risk, in particular, the effect of novel therapies. The development of more personalized risk stratification methods, potentially incorporating biomarkers and novel therapies, is needed. The next steps following this study will be to have further validation of these results with patient cohorts from other institutions, as well as the development of a risk stratification model for VTE in patients taking daratumumab-based regimens or other monoclonal antibodies for myeloma therapy. Those data could, in the future, inform clinicians to give appropriate prophylaxis for patients who may be at higher risk.

Conclusions

In our cohort of MM patients treated with daratumumab-containing regimens, we found that the occurrence of VTE was associated with the presence of medical comorbidities and the recurrent use of certain types of indwelling venous access procedures. Clinicians should look for these particular risk factors and may consider appropriate concurrent prophylaxis strategies to prevent the development of VTE.

Author contributions

N.G. Subramanian, Z.E. Hunzeker, and C.M. Rojas Hernandez performed manual chart review and wrote the paper; N.G. Subramanian, F.M. Aung, and C.M. Rojas Hernandez created the research study design; K.K. Patel and H.C. Lee edited the paper; J. Jin, E. Young, and R.L. Bassett Jr. contributed with data extraction and analyzed the results.

Funding

Analyses of this work were supported in part by the NIH/NCI Cancer Center Support Grant (award number P30 CA016672) and used the Biostatistics Resource Group (R.L.B.).

Data availability

Deidentified data is available at request, in order protect study participant privacy.

Declarations

Ethics approval and consent to participate

In accordance with the Declaration of Helsinki, the study is a retrospective evaluation of the patients and will only involve minimal risk to subjects. Complete confidentiality was maintained during this retrospective study by limiting access to the investigator and research staff. Patient Identifiers and extracted data will be stored in encrypted fashion with access to individuals with permission. The University of Texas MD Anderson Cancer Center Institutional Review Board approved the protocol and waived the informed consent for the study.

Informed consent

The need for informed consent was waived by The University of Texas MD Anderson Cancer Center Institutional Review Board.

Competing interests

C.M. Rojas Hernandez has received research support from ANTHOS therapeutics, not related to the present study.H.C. Lee has received consulting support from Bristol Myers Squibb, Celgene, Genentech, Legend Biotech, Janssen, GlaxoSmithKline, Sanofi, Pfizer, Monte Rosa Therapeutics, Oncopetides, Takeda Pharmaceuticals, Allogene Therapeutics and research support from Amgen, Bristol Myers Squibb, Janssen, GlaxoSmithSkine, Regeneron, Takeda Pharmaceuticals, not related to the present study.All other authors report no conflict of interest.