Pediatric relapsed/refractory ALK-positive anaplastic large cell lymphoma treatment and outcomes in the targeted-drug era

Anaplastic large cell lymphoma (ALCL) accounts for only 15% of all pediatric non-Hodgkin lymphomas,¹ but has a significantly higher relapse rate of 21% to 30% after frontline treatment than most other pediatric non-Hodgkin lymphomas.^2-6 Multiple clinical trials have been designed to improve event-free survival (EFS) with frontline therapy, but very few studies have focused on relapsed disease. The ALCL-Relapse trial, the first published prospective trial for pediatric relapsed or refractory (R/R) ALCL, tested a risk-stratified approach for 116 children with R/R ALCL treated at diagnosis with the ALCL99 chemotherapy backbone.^3,7 Low-risk patients received weekly vinblastine monotherapy for 24 months, intermediate-risk patients received chemotherapy and autologous hematopoietic stem cell transplant (HSCT), and high-risk and very-high-risk patients were treated with chemotherapy and allogeneic HSCT (allo-HSCT), with 5-year overall survival (OS) of 90%, 78%, 73%, and 59%, respectively. This study was performed before the introduction of targeted therapies, which have created a paradigm shift in the approach to ALCL treatment.

ALCL uniformly expresses CD30 on the cell surface, making it an ideal target for therapy. Brentuximab vedotin (BV), an antibody-drug conjugate composed of an anti-CD30 chimeric antibody conjugated to the microtubule-disrupting agent, monomethyl auristatin E, has been used in patients with relapsed ALCL achieving complete response (CR) rates of 41% to 66%.^8,9 In a phase 1/2 study of 17 pediatric patients with R/R ALCL, 53% achieved an overall response (41% CR, 12% partial response [PR]), and 13 ultimately received HSCT.⁹ In a phase 2 study of adolescents and adults with relapsed ALCL treated with BV for up to 16 cycles, 38 of 58 (66%) patients achieved a CR, and 8 of 58 (14%) patients remained in CR at 5 years without HSCT.⁸ BV is now US Food and Drug Administration (FDA) approved for the treatment of adults with newly diagnosed ALCL in combination with cyclophosphamide, doxorubicin, and prednisone.¹⁰ 

Almost all cases of pediatric ALCL are anaplastic lymphoma kinase (ALK) positive, which can also be targeted with specific therapies.¹¹ The ALK inhibitor, crizotinib (CZ), was recently FDA-approved for the treatment of relapsed ALCL based on the Children’s Oncology Group study ADVL0912 (ClinicalTrials.gov identifier: NCT00939770). In the aforementioned study, 26 patients with relapsed ALK-positive ALCL were treated with CZ, with an objective response rate of 88%.¹² Although most patients chose to come off study to go to transplant, 4 patients continued on CZ for >2 years. Additional next-generation ALK inhibitors have been developed, primarily for ALK-positive non–small cell lung cancer, and have clinical potential for treating ALCL, especially central nervous system (CNS)-positive disease, because they can effectively cross the blood–brain barrier.^13-16 

Based on early clinical trials, a phase 2 Children’s Oncology Group trial ANHL12P1 (ClinicalTrials.gov identifier: NCT01979536) evaluated the addition of BV or CZ to ALCL99 chemotherapy for de novo ALK-positive ALCL. The BV arm of the study showed a 2-year EFS of 79.1% and OS of 97.0%, whereas the CZ arm showed 2-year EFS of 76.8% and OS of 95.2%.^4,5 How the use of these targeted agents in upfront therapy may affect their efficacy in the relapse setting is unknown.

Although these promising targeted therapies have the potential to change the clinical landscape for patients with R/R ALCL, there is a lack of data on how best to integrate them into treatment. Trials of targeted agents in R/R ALCL only provide data on initial response, but long-term data on duration of response and role of HSCT is unknown. To understand how pediatric, adolescent, and young adult patients with R/R ALCL are being treated in the real-world setting in the era of targeted therapy, we conducted a retrospective, multi-institutional study evaluating the relationship between baseline disease characteristics, treatments, and clinical outcomes.

We conducted a multicenter retrospective analysis of patients diagnosed with ALCL aged ≤21 years of age at initial diagnosis and treated for R/R ALCL between 2011 and 2022. Patient data were collected from 18 academic medical centers within the United States. Centers queried their institutional databases to identify eligible patients. Patients aged ≤21 years were included if they had biopsy-proven ALCL of any stage at initial diagnosis and were treated for R/R ALCL. We chose 2011 as the starting year for this study because it was the year that both BV and CZ received FDA approval and became more available in the United States. The institutional review boards of all participating centers approved or waived the study.

Participating centers submitted data on a total of 87 patients. Study data were collected and managed using REDCap (research electronic data capture) tools hosted at Stanford University.^17,18 Deidentified data collected included: (1) demographics; (2) diagnosis and staging information; (3) type, length, and interval of treatment; (4) frequency and timing of dose modifications because of toxicity; (5) use of HSCT; (6) duration of response; and (7) outcomes. Two patients were excluded because of incomplete data and 4 patients were excluded because they were ALK-negative at diagnosis, leaving 81 patients for analysis. EFS was defined as time from relapse to progressive disease, relapse or second malignancy, or death from any cause. OS was defined as the time from relapse to death from any cause. Patients alive were censored at their last follow-up. Survival rates were estimated using a Kaplan-Meier estimator and presented with 95% confidence intervals (CIs).

A total of 81 patients with ALK-positive R/R ALCL are described. Patient characteristics at initial diagnosis are shown in Table 1. At relapse, patients had lymphoma that was Murphy stage I (8.6%), II (15%), III (38%), IV (33%), and unknown (4.9%).¹⁹ First relapse occurred a median of 8.9 months after initial diagnosis (range, 2.6-131.9), with 11 patients progressing during their initial chemotherapy, and 42 other patients relapsing within 1 year of diagnosis. Initial reinduction regimens included ALK-inhibitor monotherapy (n = 37, 46%; n = 34, CZ; n = 1, ceritinib; and n = 2, lorlatinib), BV monotherapy (n = 19, 23%), chemotherapy without targeted agent (n = 12, 15%), chemotherapy with targeted agent (n = 9, 11%), or vinblastine monotherapy (n = 4, 5%). Specific chemotherapy regimens are listed in supplemental Table 1. Overall, 67 of 81 patients (83%) achieved a CR to their initial relapse therapy. Responses by reinduction regimen are shown in Table 2. Of 4 patients who received vinblastine monotherapy, 1 achieved a CR and completed 2 years of therapy and then relapsed soon after discontinuation; 1 achieved a CR and used it as a bridge to CZ; 1 had a PR; and 1 had progressive disease. Of 25 patients who received BV as part of treatment at initial diagnosis, 4 received BV as initial relapse therapy. Of those, 2 did not respond, 1 achieved a CR and then received lorlatinib before allo-HSCT, and 1 achieved a CR but relapsed after allo-HSCT. Of 9 patients who received CZ as part of treatment at initial diagnosis, 3 received CZ as initial relapse therapy and all achieved a sustained CR. At some point in their relapse therapy, 32 patients received BV, 48 received CZ, 3 received alectinib, 1 received ceritinib, 7 received loratinib, 14 received vinblastine, and 4 received the programmed cell death protein-1 inhibitor, nivolumab. The swimmers plot in Figure 1 details the treatment courses.

Of 81 patients, 58 (71.6%) received at least 1 HSCT after relapse (11 autologous and 48 allogeneic, with 1 patient receiving both). Conditioning regimens are listed in supplemental Table 2. Fifty-six received HSCT as part of first relapse therapy and 3 after a second or subsequent relapse. There were no significant differences between the time to relapse, stage at relapse, or type of reinduction regimen used between patients who did or did not receive HSCT (Table 3). Of 47 patients, 3 did not achieve a CR before receiving allo-HSCT. One patient progressed on multiple reinduction regimens, had an unknown response before HSCT, and ultimately had progressive disease after transplant. One patient had a PR and received maintenance after HSCT and remains on therapy in remission, and 1 patient with a PR did not receive additional therapy after HSCT and is off-therapy in remission. Eight patients received maintenance therapy after HSCT, 6 received an ALK inhibitor for a median of 7.5 months (range, 1.4-22.5), and 2 received BV for a median of 5.6 months (range, 2.1-6.3). At the time of data cutoff, there were no relapses in the patients who received maintenance, with 2 patients continuing the treatment. There were 12 relapses after HSCT (4 autologous and 8 allogeneic), 2 episodes of posttransplant lymphoproliferative disorder, and 7 deaths. Of 7 deaths, 4 were due to transplant-related toxicity, 2 were due to progressive disease, and 1 was due to an unrelated cause. All but 2 of the patients who relapsed after HSCT responded to additional therapy and remain alive.

Of 23 patients who did not receive a transplant, 4 patients died and 19 were alive in remission at the end of study follow-up. Two patients never achieved a CR and died because of progressive disease; notably, neither received a targeted agent for unknown reasons. One patient achieved CR on CZ but then developed CNS disease on therapy and declined further treatment, and 1 patient died of COVID-19–related respiratory failure after achieving a CR with lorlatinib. Outcomes of patients based on the timing of their relapse and use of transplant are shown in Figure 2A.

Duration of treatment with BV or ALK inhibitors varied widely for patients who never went to HSCT or had a second relapse after HSCT (median BV, 2.3 years [range 1-11]; CZ median, 3.0 years [range, 2-9.6]; lorlatinib median, 1.5 years [range 0.5-2.4]; ceritinib, 2.9 years). Treatment for second relapse included 11 patients who received an ALK inhibitor, 4 who received BV, 1 who received vinblastine, and 2 who died before additional therapy. At the time of data collection, 19 patients remain on active treatment, 6 after allo-HSCT (2 receiving maintenance, and 4 after a subsequent relapse), 2 after auto-HSCT because of relapse, and 11 patients who never received a transplant (6 CZ, 3 BV, 1 BV + CZ, and 1 vinblastine). One patient experienced a second relapse after prior allo-HSCT and achieved a CR with CZ, then relapsed after 1.5 years on CZ and achieved a fourth CR with lorlatinib, and now continues treatment with lorlatinib, in remission for 2.4 years. One patient received BV after relapsing after auto-HSCT and after vinblastine. After achieving a fourth CR, the interval of BV was spaced out from every 3 weeks to every 9 weeks over 9 years at which point the patient relapsed. However, the patient attained a fifth CR after restarting BV every 3 weeks and continues with therapy.

Ten patients discontinued ALK-inhibitor monotherapy in remission after receiving it for relapse treatment. Three of these patients relapsed after allo-HSCT, then achieved a subsequent remission with an ALK inhibitor (treatment duration: 1 year with ceritinib and radiation therapy, 3 years with CZ, and 6.8 years with CZ) and all remain off-therapy in remission for a median of 4.5 years (range, 0.1-5.2) since stopping treatment. Seven patients who never had a transplant, 5 of whom relapsed >1 year after initial diagnosis, received treatment with CZ for a median of 2.1 years (range, 1.1-9.6) before discontinuation. One of these patients stopped CZ after 1.4 years of treatment, remained in remission off-therapy for 3.8 years and then relapsed but was able to achieve a subsequent CR with reinitiation of CZ and remains in remission on CZ 4.6 years later. The other 6 patients remain in remission for a median of 1.5 years (range, 0.3-7.1) since the end of relapse therapy.

Seven patients had CNS involvement at initial diagnosis, including 3 with positive cerebrospinal fluid (CSF), 2 with intracranial lesions, and 2 with both. These patients received CNS-directed therapy during initial treatment, and none had CNS involvement at the time of relapse. Five different patients had CNS involvement at relapse: 3 with positive CSF, 1 with intracranial lesions, and 1 with both. Another patient receiving vinblastine for isolated skin relapse developed new-onset headaches 1 month into treatment and was found to have CSF involvement with intracranial disease. Treatment used for CNS involvement included intrathecal (IT) chemotherapy in 5 patients (IT methotrexate or IT triples with methotrexate, hydrocortisone, and cytarabine), ALK inhibitor with CNS penetration in 4 patients, and radiation therapy in 1 patient, with all 6 patients achieving a CR of their CNS disease. Four patients received an allo-HSCT, with 1 dying because of treatment-related toxicity; 1 received an auto-HSCT; and 1 patient died secondary to infection before transplant. For patients who were CNS-negative at relapse, CNS prophylaxis was used in 42% of patients. This primarily consisted of IT chemotherapy given every 1 to 3 months.

For the overall cohort, 5-year EFS was 63% (95% CI, 53-75) and 5-year OS was 91% (95% CI, 84-98; Figure 2B-C). Notably, 20 patients (25%) remained on therapy at the time of data collection. There was no significant difference in EFS or OS of patients with or without bone marrow involvement at relapse; 5-year EFS was 74% (95% CI, 58-94) and OS was 90% (95% CI, 79-100) vs 58% (95% CI, 46-74) and 91% (83-100), respectively; or those with relapse <1 or >1 year from initial diagnosis, 5-year EFS was 60% (95% CI, 48-76) and OS was 89% (95% CI, 79-100) vs 69% (95% CI, 53-90) and 95% (95% CI, 87-100; supplemental Figure 1), respectively. Of those patients who achieved a CR before HSCT or achieved a CR and did not receive an HSCT, 5-year EFS was 64% (95% CI, 50-81) and OS was 94% (95% CI, 86-100) for patients with allo-HSCT, 55% (95% CI, 32-94) and 91% (95% CI, 75-100) for patients with auto-HSCT, and 68% (95% CI, 48-96) and 88% (95% CI, 73-100), respectively, for patients who did not go to transplant (Figure 2D-E).

This report is, to our knowledge, the most extensive analysis of treatment and outcomes for children diagnosed with R/R ALK-positive ALCL in the era of targeted therapy. The large number of patients across 18 academic medical centers enabled us to describe the real-world experience in managing these challenging patients. Ninety percent of the patients received either BV and/or an ALK inhibitor at some point in their relapse treatment, demonstrating how integral targeted therapies have become to the treatment of R/R ALCL. Over the timeframe captured in our retrospective study (2011-2022) there was an increase in the use of targeted therapy over time as it became more available. The ALCL-Relapse trial, the singular published prospective study of pediatric R/R ALCL (recruitment 2004-2014), did not include targeted therapy.⁷ In that study, 10 of 17 (59%; 95% CI, 40-88) very-high-risk patients, defined as those who had progressed on initial therapy, were alive 5 years after treatment with high-dose chemotherapy and allo-HSCT. In contrast, in this analysis with the addition of targeted therapy, 9 of 11 (82%; 95% CI, 59-100) very-high-risk patients who progressed on initial therapy remained alive and in remission at a median of 4.9 years (range, 1.1-10.9). CD3 was an important marker for risk stratification in the ALCL-Relapse trial but was not routinely assessed in this retrospective cohort, making it difficult to compare risk groups in the 2 studies. Notably, CR rates to reinduction treatment with an ALK inhibitor or BV on our study are higher than those reported in the initial R/R trials of BV and CZ, which is likely attributable to the fact that those trials included patients who had received multiple prior lines of treatment. ^8,9,20 

This real-world study demonstrates heterogeneity in treatment approaches and sequencing of therapies. Treatment intensity spanned high-dose chemotherapy and allo-HSCT to monotherapy with BV or an ALK inhibitor, reflecting a lack of consensus on the best treatment for R/R ALCL in the targeted-therapy era. Selection of treatment was likely influenced by previously described risk factors such as relapse <1 year from initial diagnosis, bone marrow involvement, and CNS involvement.²¹ None of those risk factors was found to be predictive in this cohort (supplemental Figure 1), although analysis was complicated by a trend toward higher risk patients (those with bone marrow and/or CNS involvement or early relapse) receiving HSCT. Notably, 30% of patients had bone marrow positivity at relapse, which is higher than previously reported, potentially because of increased sensitivity of testing. Comparison of outcomes for low-risk patients on the ALCL-Relapse trial who relapsed >1 year from initial diagnosis, were CD3⁻, and had not received prior vinblastine is limited by lack of CD3 data in our patients. However, it is notable that with 2 years of outpatient vinblastine and relatively low short- and long-term toxicity, this regimen achieved a 5-year EFS of 81% and OS of 90%.⁷ In our cohort, 17 patients who relapsed over a year after initial diagnosis received an HSCT. Although these patients also achieved good EFS and OS, it raises the question of whether similar outcomes could have been achieved without the potential toxicity of HSCT.

Although most patients who received BV or CZ as part of treatment for their initial ALCL diagnosis were treated for relapse with a targeted therapy they had not previously received, the 3 patients who received CZ at initial diagnosis and at relapse had a good response, suggesting ALK inhibitors may be a useful part of relapse therapy even when used in upfront therapy. Although most patients responded to their initial relapse therapy, the relatively low EFS of this cohort compared with the much higher OS reflects the fact that targeted therapy is effective even as second- and third-line treatment. More work needs to be done to determine the most effective sequencing and duration of therapy. Pediatric studies have demonstrated the benefit of CZ or BV in patients with R/R disease with durable remission in some patients without further therapy. However, historically, most patients who achieved a CR proceeded to HSCT.^9,12 

Uncertainty remains about remission durability after cessation of ALK inhibitors or BV monotherapies, as demonstrated by the high percentage of patients in this cohort who either continued with treatment with these agents for prolonged periods of time or underwent consolidation with HSCT. In some cases, patients achieved a remission and discontinued the medication, experienced disease recurrence, and then achieved a subsequent remission after restarting the same medication. Rapid relapse after CZ discontinuation has been reported in patients treated for R/R ALK–positive ALCL,²² even when quantitative polymerase chain reaction (PCR) of nucleophosmin-ALK in the peripheral blood had been negative for a year before CZ discontinuation.²³ Reverse transcription PCR (RT-PCR) has been shown to help track response to therapy, with 1 single-center study of 25 adolescent and adult patients with R/R ALK-positive ALCL demonstrating that failure to achieve a negative RT-PCR while on treatment with CZ was associated with risk of disease progression.²⁴ The same study reported on the longest known follow-up of patients treated with an extended duration of CZ, median of 5.1 years (range, 0.5-11.7), which was well tolerated overall in the adult population. Eight patients in our retrospective cohort received >3 years of CZ as part of their therapy, which was well tolerated. Unfortunately, RT-PCR monitoring of nucleophosmin-ALK is not currently available in the United States from a Clinical Laboratory Improvement Amendments–certified laboratory, limiting the ability to incorporate that data into clinical decision-making. More information on the long-term side effects of ALK inhibitors is needed to inform treatment decisions in pediatric patients.

Although CZ was the most common ALK inhibitor used for treatment in this retrospective cohort, second- and third-generation ALK inhibitors are increasingly being used in the treatment of R/R ALK-positive ALCL. Reasons behind choice of medication range from relapse on CZ or presence of CNS involvement to insurance approval. Alectinib and ceritinib, second-generation ALK inhibitors, have been studied in a phase 1 trials of R/R ALK-positive ALCL with overall response rates of 80% and 75%, respectively.^14,15 Another second-generation ALK inhibitor, brigatinib, has demonstrated efficacy in treatment of CZ-resistant ALK-positive ALCL in patient-derived xenograft mouse models.²⁵ Brigatinib is currently being studied in a phase 1/2 trial for adults and children with R/R ALK-positive ALCL (ClinicalTrials.gov identifier: NCT04925609).

There is currently no consensus regarding the best approach for treating patients with CNS involvement at diagnosis or relapse. In the ALCL99 trial, patients with CNS involvement represented 2.6% of registered patients (12/463) but were excluded from participation and had worse outcomes with 5-year EFS of 50% and OS of 74%.²⁶ In a review of 618 patients treated with ALCL99 therapy, 26 pediatric patients (4.2%) had CNS involvement at first or subsequent relapse at a median follow-up of 8 years.²⁷ Treatment approach for CNS relapse was heterogenous, and 3-year OS after CNS relapse was 48.7%. CNS progression during relapse treatment for patients with high-risk R/R ALCL who are initially CNS negative has been described for patients receiving relapse treatment with vinblastine, CZ, or BV, all of which have limited CNS penetration.²⁸ With the introduction of second- and third-generation ALK inhibitors, treatment options for CNS disease are improving. A case series of 10 patients with 11 episodes of CNS relapse/progression of ALK-positive R/R ALCL treated with second- or third-generation ALK inhibitors reported 10 CRs.¹⁶ Four patients in our cohort had CNS disease treated with an ALK inhibitor with CNS penetration, whereas the others were treated with IT chemotherapy and/or radiation therapy with good response. Next-generation ALK inhibitors provide effective oral therapy for ALK-positive CNS lesions and are less invasive with potentially fewer long-term side effects than prior treatment options.

Another promising treatment for R/R ALCL is immune checkpoint inhibition because programmed death-ligand 1 is expressed in ALK-positive ALCL and the immune response plays an important role in disease control.²⁹ Case reports have demonstrated good responses to the programmed cell death protein-1 inhibitors nivolumab or pembrolizumab in treatment of patients with R/R ALK-positive ALCL.^30-32 Four patients in our study received nivolumab at some point in therapy. Further information is needed on the efficacy of checkpoint inhibitors in this patient population, which is currently being studied in the NIVOALCL trial (ClinicalTrials.gov identifier: NCT03703050).

Several limitations should be considered when interpreting our findings. First, although we tried to assemble a representative cohort of patients across multiple institutions based on inclusion of all eligible patients treated for R/R ALK-positive ALCL in the given timeframe, it is possible that there was selection bias, and some patients were missed or had incomplete data given the retrospective nature of the study. Second, we were unable to capture the rationale for why patients did not receive HSCT. Possible explanations include inability to achieve CR, prior toxicities, poor donor options, and family and patient preference. Although patients who received HSCT were more likely to be off-therapy and alive at the time of data cutoff, there were 4 deaths due to transplant-related toxicity and no deaths related to treatment toxicity in the group who did not receive HSCT. Notably, patients were more likely to relapse after HSCT than if they continued with ALK inhibitor or BV therapy while in CR, although patients who did not receive HSCT had much longer treatment duration. The optimal duration of targeted therapy remains unknown. This raises the question of how to weigh the short- and long-term toxicities of allo-HSCT against the less well–defined toxicities of prolonged treatment with an ALK inhibitor or BV.

In summary, this first real-world report of the treatment of pediatric, adolescent, and young adult patients with R/R ALK-positive ALCL highlights the excellent responses achieved with ALK inhibitor or BV monotherapy and demonstrates that HSCT may not be necessary for all R/R patients to achieve durable remission. However, questions persist about how best to identify patients who may benefit from HSCT, the optimal duration of targeted therapy, and the use of combination therapy necessary to achieve a cure. Future prospective studies are needed to determine how best to incorporate targeted therapies into treatment of relapsed ALK-positive ALCL.

This work was supported by the Stanford University Maternal and Child Health Clinician Educator Grant. The Stanford REDCap platform is developed and operated by Stanford Medicine Research Technology team. The REDCap platform services at Stanford are subsidized by Stanford School of Medicine Research Office, and the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through grant UL1 TR003142.

Contribution: L.J.M., L.M.S., E.J.L., and C.A. conceived of and designed the study; L.J.M., J.E.A., K.Y.K., J.R., R.G., M.J.E., K.J.D., C.A.P., A.R., K.A., J.W., P.S., C.J.F., C.M.S., C.G., Z.A., C.H.L., J.A.B., H.D., D.H., K.T., M.P.L., L.M.S., E.J.L., and C.A. provided patients and collected data; V.R. performed the statistical analysis; L.J.M. drafted the manuscript; and all authors participated in the analysis and interpretation of data and critical revision of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lianna J. Marks, Division of Pediatric Hematology, Oncology, Stem Cell Transplantation and Regenerative Medicine, Department of Pediatrics, Stanford University School of Medicine, 750 Welch Rd, Suite 200, Stanford, CA 94304; email: marksl@stanford.edu.