Chimaeric anti-CD20 monoclonal antibody (rituximab) in post-transplant B-lymphoproliferative disorder following stem cell transplantation in children

Authors


Dr Albert Faye, Service d'hémato-immunologie, Hôpital Robert Debré, 75019 Paris, France. E-mail: albert.faye@ rdb.ap-hop-paris.fr

Abstract

Post-transplant lymphoproliferative disorder (PTLD) after haemopoietic stem cell transplantation is a serious complication that occurs in 8–22% of patients with high-risk factors. We retrospectively investigated tolerance and efficacy of humanized anti-CD20 monoclonal antibody (rituximab) as first-line treatment in 12 children with B-cell PTLD. At diagnosis, eight patients had tumoral involvement. The other four patients had fever, associated with raised Epstein–Barr virus (EBV) viral load and monoclonal gammopathy. Rituximab was given at the dose of 375 mg/m2 once a week by intravenous infusion (1–9 infusions). Only 1/48 infusions was associated with a grade 2 clinical adverse event. Eight out of 12 (66%) patients responded to the treatment and were in complete remission. All patients without tumoral involvement responded to the treatment. A rapid decrease in fever within 1 week was observed in all responders. Non-responders did not show any clinical response during the first week. Tumoral involvement and immunodepression seemed to be more marked in non-responders. Rituximab was an effective and well-tolerated treatment of B-cell PTLD. Early treatment before tumoral involvement seemed to be the most effective approach. Lack of rapid response should lead to intensification of PTLD treatment. Pre-emptive treatment should be considered and evaluated in further longitudinal multicentre studies.

Post-transplant lymphoproliferative disorder (PTLD) is an uncommon but severe complication of stem cell transplantation (SCT). Generally it is B-lymphocyte proliferation induced by Epstein–Barr Virus (EBV) in severely immunocompromised patients. B-cell PTLD occurs in about 1% of SCT but the frequency may increase up to 8–22% in high-risk patients (Zutter et al, 1988; Sociéet al, 2000). Major risk factors are unrelated or human leucocyte antigen (HLA)-mismatched related SCT, T-cell depletion and use of antithymocyte globulin (ATG) for acute graft-versus-host disease (GVHD) prophylaxis (Curtis et al, 1999; Sociéet al, 2000). Prognosis is poor, particularly after SCT, resulting in rates up to 90% mortality (Shapiro et al, 1988; Fischer et al, 1994).

There are only a few therapeutic approaches to B-cell PTLD following SCT. Reduction of immunosuppression can result in up to 50% of patients reaching remission in B-cell PTLD following organ transplant (Starzl et al, 1984). However reduction of immunosuppression in PTLD following SCT may not be effective, as patients are severely immunocompromised.

Adoptive immunotherapy using donor lymphocytes infusion (DLI) is an effective treatment, but is sometimes complicated by GVHD and/or respiratory distress syndrome (Papadopoulos et al, 1994). Infusion of specific anti-EBV cytotoxic T lymphocytes (CTL) is an alternative treatment, but this procedure is not available in most SCT centres (Rooney et al, 1995). The use of monoclonal anti-B-cell antibodies is a specific novel therapeutic approach that has shown promise. Treatment of B-cell PTLD with monoclonal anti-CD21 and anti-CD24 murine antibodies has led to 57% of B-PTLD remission in 28 patients who had undergone SCT (Benkerrou et al, 1998). Unfortunately these antibodies are no longer available. Moreover, it is hypothesized that the efficacy of such treatment is decreased by the production of human anti-murine antibodies.

Rituximab is a highly specific mouse/human chimaeric anti-CD20 antibody. It involves variable murine regions targeting the CD20 B-lymphocyte antigen and human IgG1 heavy chain and light chain (kappa) constant regions (Maloney et al, 1996). Rituximab has been approved for the treatment of relapse of low-grade or follicular non-Hodgkin's lymphoma (Maloney et al, 1997).

We have previously described prolonged complete remission of B-cell PTLD following rituximab treatment in a child after matched unrelated SCT (Faye et al, 1998). Other preliminary reports have shown the efficacy of such treatment in B-cell PTLD after organ transplantation or SCT (Kuehnle et al, 2000; Milpied et al, 2000). However, these rare data concern very few patients and children.

In the present retrospective study we report the use of rituximab for B-cell PTLD following SCT in 12 children. We describe the course of the disease and the features of responders and non-responders. We show that rituximab is an effective and well-tolerated treatment in most of these patients, but that some factors may predict treatment failure.

Patients and methods

Patients

The study was conducted in five French paediatric SCT centres: Hôpital R. Debré, Hôpital Necker-Enfants Malades, Hôpital Saint Louis, Paris, CHU de Strasbourg and CHU de Nancy. Between June 1997 and December 1999, 187 matched unrelated or mismatched related SCT were performed in these centres. Twelve children (6·4%) developed a B-PTLD. All of them received rituximab as first-line treatment. Characteristics of the patients are presented in Table I. Median age at SCT was 5 years (range 11 months to 16 years). Indications for SCT were haematological malignancies (n = 6), Fanconi syndrome (n = 3), adrenoleucodystrophy (n = 1), metabolic disease (n = 1) and congenital immunodeficiency (n = 1).

Table I.   Patient characteristics.
PatientUnderlying
disease
Age at SCT
years/months
EBV status
Donor/recipient
Type of
SCT
Conditioning
regimen
GVHD
prophylaxis
  • *

    ATG (thymoglobuline) was given at the total dose of 15 mg/kg divided into three injections at d −3, −2 and −1 (patients 1, 3 and 4) and divided into five doses at d −5, −4, −3, −2 and −1 (patients 8, 9 and 10). Patient 7 received 30 mg/kg divided into three doses at d −3, −2 and −1. Patient 2 received 20 mg/kg divided into five doses (d −5, −4, −3, −2 and −1) and patient 12 received 11 mg/kg divided into three doses (d −3, −2 and −1).

  • †T-cell depletion. After T-cell depletion, the number of CD3+ cells × 104/kg infused was 3·0 (patient 5), 6·8 (patient 7), 3·6 (patient 8), 0·15 (patient 10) and 0·28 (patient 11).

  • CID, combined immunodeficiency; MUD, matched unrelated donor; Haplo-id, haplo-identical; Fluda, fludarabine; Me, melphalan; TAM, TBI + aracytine + melphalan; CY, cyclophosphamide; Bu, busulphan; Csa, cyclosporine; MTX, methotrexate (administered at d 1, 3 and 6 post SCT except for patients 1 and 3 (d 1 and 3 post SCT only); Pred, prednisone.

1AML2/0+/–MUDFluda-Me-ATG*MTX-Csa-Pred
2Fanconi14/0+/+MUDTBI-CY-ATGCsa-Pred
3ALL3/6+/+MUDTAM-ATGMTX-Csa-Pred
4AREBT12/10+/–MUDTAM-ATGMTX-Csa
5Ghünter2/0+/–Haplo-idBu-CYNo
6ALD6/8+/+MUDBu-CYCsa
7T-NHL14/3+/+Haplo-idTAM-ATGNo
8Fanconi12/0+/+MUDTBI-CY-ATGCsa
9AML1/4+/+MUDBu-CY-ATGMTX-Csa
10Fanconi0/11+/+MUDTBI-CY-ATGCsa
11CID1/2+/+Haplo-idBu-CYNo
12AML16/5+/+MUDTAM-ATGMTX-Csa-Pred

Risk factors for PTLD were matched unrelated SCT for nine children and mismatched related SCT for three patients. Nine patients received ATG (rabbit ATG, thymoglobuline) within the conditioning regimen for GVHD prophylaxis and the graft was T-depleted in five cases (patients 5, 7, 8, 10 and 11) using a method of positive selection of CD34-positive cells (Clinimax, Ceprate, Baxter or Mylteni). The number of T cells infused for these patients is presented in Table I. The number of CD34+ cells infused was 2·1 × 106 (patient 8) to 20·4 × 106 cells/kg (patient 3) (median: 4·19 × 106 cells/kg). All patients received a bone marrow graft except one (patient 7) who received a mobilized blood stem cell graft. Complete donor chimaerism was obtained in all but one patient (patient 6). Irradiation was included in the conditioning regimen for seven patients. Three children had GVHD greater than grade I after SCT (patients 8, 9 and 12). Patient 8 presented GVHD grade II, according to the Glucksberg classification (Glucksberg et al, 1974), at d 18 post-SCT that necessitated methylprednisolone up to 5 mg/kg/d, cyclosporine and tacrolimus. Patient 9 presented grade III GVHD at d 14 and 69 post-SCT, which was treated with 5 mg/kg/d methylprednisolone, cyclosporine and ATG (lymphoglobuline and thymoglobuline). Patient 12 presented at d 28 post-SCT with GVHD grade III, which was treated with methylprednisolone 5 mg/kg/d, cyclosporine, anti-interleukin 2 (IL-2) receptor monoclonal antibodies (Leucotac) from d 47 to d 74 post-SCT and mycophenolate.

Diagnosis of B-cell PTLD

PTLD was diagnosed on the basis of a finding of diffuse hyperplasia characterized by invasion of blood vessels and other organ structures, with disorganization of the nodal structure in lymph nodes. If pathological investigation was not possible, PTLD was diagnosed by a raised EBV viral load, fever and lymph node enlargement. Tumours were classified morphologically according to Nalesnik et al (1988) as polymorphic or monomorphic PTLD by a local pathologist.

General presentation (Table II) B-PTLD occurred at a median of 77 d following SCT (range 29–205 d). All patients had fever at diagnosis. Eight patients had either clinical enlargement of lymph nodes, hepatomegaly, splenomegaly, tonsil or cavum localization. No patient had central nervous system involvement.

Table II.   B-PTLD characteristics at diagnosis.
PatientTime SCT/
B-PTLD (d)
LocalizationHaemophagocytic
syndrome
EBV viral load
(eq. copies/105 cells)
CD4 cell
count (μl)
B-cell
count (μl)
  1. LN, lymph nodes; NE, not evaluable; NA, not available.

194Cervical and intra-abdominal LN,Yes10 000147472
tonsil, splenomegaly    
2205Cervical LN, tonsilNo 400022317
396Cervical and mediastinal LN,Yes10 0009108
splenomegaly, hepatomegaly    
460Cervical and intra-abdominal LN,NEEBV PCR positive49120
tonsil, splenomegaly    
530Splenomegaly, hepatomegalyYes20 0004080
629Cervical and mediastinal LNNo200 0000200
cavum, splenomegaly, hepatomegaly    
758No 3000050
8131No30 00049NA
9117No30 000263NA
1036No30 0002140
1146HepatomegalyNo 30002727
1266Cervical and mediastinal LN,No10 00000
lung, kidney    

Diagnostic criteria Pathological documentation was available for three patients (patients 2, 4 and 12). Histology showed typical B lymphoproliferation with positive EBV in situ hybridization using an EBER (EBV-encoded small RNA) probe. Monoclonality of two cell proliferations was assessed and confirmed by study of IgH rearrangement with Southern blotting using a probe for the sequence encoding the heavy-chain joining region. Cell proliferation in patient 12 was oligoclonal as assessed by immunoglobulin expression detected by immunofluorescence. Donor origin of B-cell proliferation was not assessed.

B-cell PTLB was diagnosed in four patients (patients 7–10) by the association of fever, raised EBV viral load and monoclonal gammopathy.

Three patients (patients 1, 3 and 5) had haemophagocytic syndrome at diagnosis.

The characteristics of PTLD at diagnosis are presented in Table II.

Immunosuppression and immunological status at diagnosis

Ongoing treatment at B-PTLD diagnosis did not include immunosuppressive drugs for five patients (patients 2, 3, 5, 7 and 11). Other patients were receiving cyclosporine and/or steroids. Following diagnosis of PTLD, immunosuppressive treatment was stopped or decreased for all patients except one who had corticosteroid-resistant GVHD and who was receiving methylprednisolone 2·5 mg/kg/d (patient 12). In patient 1, cyclosporine was stopped and prednisone reduced from 0·5 to 0·25 mg/kg/d, 6 d before rituximab. In patients 4 and 6 cyclosporine was stopped, respectively, 5 and 8 d before rituximab. For patient 8 prednisone was decreased from 0·3 to 0·1 mg/kg/d at initiation of rituximab. In patient 9 cyclosporine was stopped 4 d after initiation of rituximab, and prednisone was maintained at 1·4 mg/kg/d.

Of interest, within the month prior to the diagnosis of PTLD no patient had a CD4 cell count of more than 100/μl, despite the fact that a few children had more than 100 CD4 cells/μl at diagnosis of PTLD (patients 1, 9 and 10). Within the first month of SCT, severe herpes simplex virus (HSV) infection occurred in two patients (patients 7 and 12) and digestive adenovirus infection occurred in patient 3. Two patients presented a cytomegalovirus (CMV) infection (patients 2 and 3), respectively, at d 66 and 60 post-SCT and patient 2 presented with a cutaneous aspergillosis infection at d 120 post-SCT. At diagnosis of PTLD, the median CD4 cell count/μl was 33 (range 0–263/μl).

Epstein–Barr virus detection and monoclonal gammopathy

EBV viral load was measured, except for patient 4, using a semiquantitative polymerase chain reaction of the EBV genome after separation of peripheral blood mononuclear cells. At PTLD diagnosis, median EBV viral load was 10 000 equivalent (eq) copies/105 cells (range 3000–200 000 copies/105 cells). Detection of EBV DNA by polymerase chain reaction was positive for patient 4.

All patients had monoclonal gammopathy evaluated by immunoelectrophoresis.

Treatment with rituximab

Rituximab was given as first-line treatment at the dose of 375 mg/m2 once a week by intravenous infusion according to the manufacturer's standard protocol (Produits Roche). In this retrospective study the number of infusions ranged from 1 to 9.

Treatment assessment

Complete remission was defined as complete disappearance of clinical or radiological tumour at all initially involved sites and the absence of new involved sites. For patients without any tumour at diagnosis it was defined by the disappearance of clinical symptoms such as fever and disappearance of monoclonal gammopathy. Partial remission was defined as at least 50% tumour size reduction. Follow-up data were collected up to 1 December 2000.

Results

Treatment with rituximab

The median interval between first symptoms and first infusion of rituximab was 6 d (range 3–18 d). Infusions of rituximab were well tolerated by all patients except patient 8, who developed fever and shivering at the first dose. The infusion time was extended and the patient subsequently received 8-weekly infusions without any side-effects.

For all patients, the median duration of infusion was 4 h (range 2–20 h). The number of infusions and interval between first symptoms of PTLD and first infusion of rituximab are presented in Table III.

Table III.   Characteristics of treatment with rituximab.


Patient
Interval between
first symptoms
and treatment (d)

Number of
infusions


Side-effects
164No
234No
362No
484No
5183No
693No
7104No
839Yes
1st infusion
938No
1064No
1131No
12112No

Responders to rituximab

Eight out of 12 patients (66%) responded to the treatment. These patients achieved complete remission (CR) after a median of 25 d following the first infusion (range 6–60 d), and included four patients who had fever, raised EBV viral load and monoclonal gammopathy without any evidence of tumour at diagnosis. Among the remaining four responders with tumoral involvement, two had monoclonal cell proliferation assessed by pathology of the tumour (patients 2 and 4). None of the responders had mediastinal involvement. Interestingly, all the responders showed signs of clinical improvement within 1 week.

The median EBV viral load was 20 000 copies/105 cells. Semi-quantitative polymerase chain reaction of EBV DNA showed that the viral load had decreased to undetectable levels for all these patients after a median of 48 d (range 6–232 d).

All patients but one were alive in CR with a median follow up of 23·5 months after CR (range 12–34 months). One patient (patient 7) died of staphylococcal sepsis as a result of central venous catheter infection 12 d after CR. Patient 1 received a DLI on d 29 and 56 after CR and patient 2 received a DLI on d 20 after CR. These two patients developed GVHD grade I following DLI. The characteristics of responders are presented in Table IV.

Table IV.   Characteristics of responders and non-responders.
CharacteristicsResponders (n = 8)
patients 1, 2, 4, 5, 7, 8, 9 and 10 (range)
Non-responders (n = 4)
patients 3, 6, 11 and 12 (range)
Median interval SCT/PTLD (d)77 (30–128)56 (29–66)
Median number of localizations1 (0–4)3·5 (1–5)
(four without patient 11)
Mediastinal localization0/83/4
Median EBV viral load20 000 (3000–30 000)10,000 (3000–200 000)
eq. copies/105 cells
Median CD4 cell count/μl49 (0–263)4 (0–27)
Median B cell count/μl100 (0–472)65·5 (0–200)
Clinical response within 1 week8/80/4

Non-responders to rituximab

Four out of 12 patients did not respond to the treatment. All these patients had evidence of tumour at diagnosis, three of them with a mediastinal localization. None of these patients showed any reduction in fever following rituximab. The median EBV viral load was 10 000 copies/105 cells. Two out of three patients (patients 3 and 12) did not show a decrease in EBV DNA viral load following rituximab treatment. The EBV viral load of these two patients was 10 000 copies/105 cells before treatment and remained stable until DLI for patient 3 (d 9 post rituximab) and death for patient 12 (d 11 post rituximab). Patient 6 experienced a dramatic decrease of viral load from 30 000 copies/105 cells before treatment to 30 copies/105 cells at d 4 and 300 copies at d 10 post treatment, despite tumoral progression. Three of the four non-responders died without any response despite the use of polychemotherapy, anti-IL6 treatment or local irradiation. Patient 3 achieved CR 32 d after DLI given 9 d after the first infusion of rituximab. In this patient DLI was associated with severe respiratory distress syndrome and GVHD grade I. This patient is still alive 19 months after CR. The other three patients died of B-PTLD progression within 2 months. For patient 11, at initiation of rituximab, only 6% of B peripheral lymphocytes were CD20 positive. At death, all B peripheral lymphocytes were CD20 negative. For patients 3 and 11, all peripheral B lymphocytes were CD20 positive. For patient 12, the pathology of the tumour showed CD20-positive B lymphocyte proliferation. Characteristics of non-responders are presented in Table IV.

Discussion

Few data are available concerning the use of rituximab in children who develop PTLD after SCT. To our knowledge, this retrospective study of 12 children is the largest one reported to date. We describe the characteristics of responders and non-responders.

Tolerance of rituximab was good in nearly all children. Benkerrou et al (1998) reported that about one third of patients experienced clinical side-effects less than grade III with anti-CD21 and anti-CD24 monoclonal antibody. In our study, a total of 48 infusions were given of which only one was associated with a grade II fever. Rituximab-related neutropenia following infusion was not observed in our study whereas it has been reported in 42% of patients treated with anti-CD21 anti-CD24 monoclonal antibodies. As rituximab is humanized it can be hypothesized that it is tolerated better than anti-CD21 or anti-CD24 monoclonal murine antibody. Cytokine release syndrome and tumoral lysis syndrome have been described in 10% of patients treated with rituximab for non-Hodgkin's lymphoma relapse. We did not observe such a complication, almost certainly because of the lack of patients with very large tumoral mass.

Rituximab is known to induce B-cell depletion for up to 6–9 months post treatment and to decrease immunoglobulin levels (Maloney et al, 1997). In our severely immunocompromised patients it was difficult to evaluate this parameter. Kuenhle et al (2000) estimated that rituximab had induced profound B-cell depletion for 7 months or more without any increase of opportunistic infections in three patients with post-SCT lymphoproliferative disorder. In our study one patient died in CR from a severe infection 1 month after the start of rituximab treatment. However, we were not able to relate such infection to the use of rituximab as the patients in our study were at high risk of infectious complications following SCT.

We considered rituximab to be an effective treatment as 66% of children were responders. Because of the small number of children involved, these data cannot be compared with the results of other studies using monoclonal antibodies. However, Benkerrou et al (1998) reported that anti-CD21 and anti-CD24 treatment induced complete remission in patients with PTLD following SCT for haematological malignancy in 28% of cases (3/11) versus 76% (13/17) for congenital immunodeficiency. In our study four out of six patients (66%) treated for PTLD after SCT for haematological malignancies achieved CR. Monoclonal B-cell PTLD is also considered to carry a poor prognosis (Malatack et al, 1991). In our study two patients had proven monoclonal proliferation that responded to rituximab.

Although the number of patients in our study was small, responders and non-responders showed different features. All responders showed rapid improvement of general signs such as decreased fever within 1 week after the first infusion. Non-responders did not improve at all. This suggests that early response to rituximab could be predictive of its efficacy. If there is no early response, the therapeutic strategy should be reviewed and may be switched to donor lymphocyte infusion, for example.

The median number of tumour sites involved was one in responders and 3·5 in non-responders. This is in agreement with the findings of Benkerrou et al (1998) in which multivisceral disease was a predictive factor of failure of anti-CD21 anti-CD24 monoclonal antibodies. Moreover, in our study three out of four non-responders had a mediastinal localization, but not responders. This localization probably reflects the severity of tumoral involvement.

It has been recently shown that decrease of EBV viral load following the use of rituximab does not seem to predict response to the treatment (Yang et al, 2000). Yang et al (2000) observed a dramatic fall in EBV viral load in three patients treated with rituximab for a PTLD after organ transplantation, despite tumoral progression. This observation and restricted expression of EBV-latency antigens in peripheral blood mononuclear cells suggested that virally infected cells in peripheral blood belong to a separate compartment from tumour cells. In our study, only one out of three non-responders experienced a decrease of viral load despite clinical progression. Although the number of patients was low, it can be hypothesized that peripheral B lymphocytes are rather tumoral in PTLD following SCT, as the patients are severely immunocompromised compared with the patients with PTLD following organ transplantation. However, in our study expression of latency EBV antigens in B lymphocytes was not done although it may be interesting to analyse this, as in Yang et al (2000).

Interestingly, in our study all patients with raised EBV viral load, gammopathy and fever without tumoral involvement responded to the treatment. This suggests that rituximab is highly effective in situations of early onset B-cell PTLD. A pre-emptive strategy before tumoral involvement could be the most suitable one and could be based on the predictive factors of B-cell PTLD, such as raised EBV viral load and monoclonal gammopathy.

Several mechanisms of failure to rituximab can be discussed. In one non-responder, we observed a decrease in CD20-positive cells in peripheral blood. It could be of interest to study the regulation of expression of CD20 on B lymphocytes by rituximab in order to analyse potential mechanisms of CD20 downregulation that could lead to rituximab failure. The median CD4 cell count was 4/μl in non-responders whose counts were all < 27 CD4 cells/μl, whereas the median CD4 cell count of the responders was 49/μl. Although the number of patients in each group was low, it can be hypothesized that rituximab could be more efficient in patients who are not too immunocompromised after SCT, as one of the mechanisms of action of rituximab is an antibody-dependent cell-mediated cytotoxicity mediated by large granular lymphocytes. Further studies of residual immunity and especially anti-EBV CTL using a tetramer approach could be of interest to assess predictive factors of treatment success.

Use of rituximab in B-cell PTLD has certainly changed the prognosis of this complication in severely immunocompromised patients. Our study demonstrates that pre-emptive use of this treatment should be considered. A knowledge of predictive factors of rituximab failure or success is needed in order to improve the management of these patients and to permit a rapid switch to a more complex cell therapy approach if failure is predicted. Further multicentre studies are now needed to evaluate these predictive factors better. Finally, combinations of monoclonal anti-B cell antibodies should be investigated as future therapeutic prospects that could decrease the failure of anti-B cell treatment.

Ancillary