Extracorporeal photochemotherapy for paediatric patients with graft-versus-host disease after haematopoietic stem cell transplantation

Authors


  • *

    Present address: R. Dall'Amico, Division of Paediatrics, ULSS 4 “Alto Vicentino”, Thiene, Italy.

Dr Chiara Messina, Oncoematologia Pediatrica, Università di Padova, Via Giustiniani 3, 35128 Padova, Italy. E-mail: chiara.messina@unipd.it

Abstract

Summary. This study aimed to ascertain whether extracorporeal photochemotherapy (ECP) is an effective treatment for paediatric patients with refractory graft-versus-host disease (GVHD). From January 1992 to December 2000, 77 children (median age 8·6 years) with either acute (n = 33) or chronic (n = 44) GVHD, resistant to conventional immunosuppressive therapy, were treated with ECP in four Italian paediatric hospitals. After ECP, acute GVHD involving skin, liver and gut responded completely in 76%, 60% and 75% of patients respectively. The 5-year overall survival was 69% for responding patients vs 12% for non-responders (P = 0·001). Among the 44 children with chronic GVHD, 15 (44%) showed a complete response and 10 (29%) a significant improvement after ECP. The 5-year overall survival was 96% for responders vs 58% for non-responders (P = 0·04). Our results suggest that ECP is an effective treatment that may be useful in paediatric patients with either acute or chronic GVHD who have failed to respond to standard immunosuppressive therapy.

Up to 30% of human leucocyte antigen (HLA)-identical sibling and 80% of unrelated donor haematopoietic stem cell transplant (HSCT) recipients develop significant graft-versus-host disease (GVHD) (Ringden et al, 1986; Storb et al, 1986; Ferrara & Deeg, 1991; Sullivan et al, 1991; Nash et al, 1992, 2000; Blazar et al, 1997; Pavletic et al, 1999). Glucocorticoids remain the standard front-line approach for patients with acute GVHD (aGVHD), whereas various therapies have been employed for patients with chronic GVHD (cGVHD)(Zecca & Locatelli, 2000). Several studies have proved that prognosis is poor for patients with aGVHD failing to respond to front-line treatment with steroids, and for patients with extensive cGVHD who do not reach satisfactory disease control (Sullivan et al, 1981; Deeg et al, 1985; Deeg & Hensley-Downey, 1990; Weisdorf et al, 1990; Martin et al, 1991; Locatelli et al, 1993; Anasetti et al, 1994; Hings et al, 1994; Hebart et al, 1995; Van Lint et al, 1998; Przepiorka et al, 2000; Khoury et al, 2001).

Extracorporeal photochemotherapy (ECP) has been proposed as a therapy for GVHD that is resistant to conventional therapy (Edelson, 1989; Lazarus & Rowe, 1995; Lazarus et al, 1997; Knobler, 2000), in view of its effectiveness in autoimmune diseases, as well as in the control of rejection after solid organ transplantation (Edelson et al, 1987; Rook et al, 1990, 1999; Costanzo-Nordin et al, 1992; Knobler et al, 1992; Dall'Amico et al, 1995; Barr et al, 1998). A few studies, enlisting a limited number of children, have reported on the use of ECP for treatment of resistant GVHD in paediatric patients, and these studies have reported promising results (Rossetti et al, 1995; Konstantinow et al, 1996; Dall'Amico et al, 1997; Dall'Amico & Zacchello, 1998; Salvaneschi et al, 2001). In this multicentre, retrospective study, we analysed the safety and efficacy of ECP in the largest cohort of paediatric patients reported to date, who had either steroid-resistant aGVHD or cGVHD resistant to two or more immunosuppressive treatment modalities.

Patients and methods

From January 1992 to December 2000, 77 patients (56 males, 21 females) with resistant GVHD were given ECP at four centres (Padova, n = 33; Pavia, n = 24; Genova, n = 15; Monza, n = 5) affiliated to the Italian Association for Paediatric Haematology and Oncology (AIEOP). Patients had been transplanted in nine AIEOP HSCT units (Brescia, Genova, Monza, Pavia, Padova, Pesaro, Pescara, Torino, Trieste). All patients had received fluconazole from d − 7 and high-dose immunoglobulin (400 mg/kg every week from d − 7 to d 100) for infection prophylaxis. For prevention of cytomegalovirus (CMV) and herpes virus reactivation, acyclovir (500 mg/m2 every 8 h intravenously) was administered from d − 7 to d 30 and then orally for 6 months after transplant. For prophylaxis of Pneumocystis carinii pneumonia, oral cotrimoxazole, starting on d + 30 and continuing until the end of immunosuppressive therapy, was administered. The expression of pp65 human CMV matrix protein on the surface of peripheral blood leucocytes was monitored in all patients to detect CMV reactivation. Patients experiencing CMV reactivation were usually treated with ganciclovir or foscarnet. Cyclosporin A, at 1–3 mg/kg/d intravenously from d − 1 to d 30, and then orally at 6 mg/kg for 6–12 months (depending on the original disorder), was administered in children who received HSCT from an HLA-identical sibling. (Locatelli et al, 2001). Either short-term methotrexate plus 3·75 mg/kg rabbit antithymocyte globulin (ATG) on d − 4 and − 3 or steroids was combined with cyclosporine for GVHD prophylaxis of patients undergoing unrelated HSCT and unrelated cord blood transplantation respectively. The median age at transplant was 8·6 years (range 0·3–20·5 years). Further details on patient characteristics are given in Table I. Informed consent was obtained from either the patient or their parents, whichever was applicable, and each Institution's Ethical Committee approved the use of ECP for treatment of GVHD.

Table I.  Clinical characteristics of the patients who underwent ECP for resistant aGVHD or cGVHD.
 aGVHDcGVHD
  1. ALL, acute lymphoblastic leukaemia; AML, acute myeloid leukaemia; CML, chronic myeloid leukaemia; BMT, bone marrow transplantation; PBSCT, peripheral blood stem cell transplantation; CBT, cord blood transplantation; MUD, matched unrelated donor; TBI, total body irradiation.

Patient3344
Median age at ECP, years (range)9·6 (1·5–18·3)8·2 (0·3–20·5)
Sex (M/F)28/528/16
Median body weight at ECP, kg (range)30 (10–85)34·5 (15–68)
Disease
 ALL1718
 AML79
 CML35
 NHL1 
 Thalassaemia major6
 Other non-malignant diseases56
Source of haematopoietic stem cells
 HLA-identical sibling BMT521
 HLA-identical sibling PBSCT 2
 MUD BMT2114
 MUD PBSCT1 
 MUD CBT2 
 HLA-identical relative BMT36
 HLA-identical relative PBSCT1 
 Haploidentical family donor 1
Conditioning regimen
 TBI yes/no9/2424/20
Median interval GVHD-ECP (range)30 d (5–91 d)8·9 months (0·4–109 months)
Median interval BMT-ECP (range)45 d (13–98 d)1·5 years (0·3–9·2 years)
Symptoms
 Skin3336
 Mouth 26
 Liver1520
 Lung 14
 Gut2021
 Joints 14
Therapies pre-ECP
 Steroids3343
 Cyclosporin A3343
 Azathioprine 4
 Okt3 2
 Methotrexate53
 Puva 3
 Thalidomide 11
 Mofetil mycophenolate37
 Fk50626

Clinical assessment

The clinical stage of organ involvement was graded and then combined to obtain an overall grade, according to published criteria, for either aGVHD or cGVHD (Ferrara & Deeg, 1991; Sullivan et al, 1991; Przepiorka et al, 1995).

Histological confirmation was obtained whenever clinically indicated to confirm GVHD diagnosis.

In the patients with cGHVD, skin involvement was assessed by an experienced paediatric rheumatologist, as described elsewhere (Kahaleh et al, 1990). The following skin parameters were evaluated at baseline and at monthly intervals: (1) extent of skin surface involved (ESS), recorded on a body diagram used to calculate surface burns in children, with scores ranging from 0% to 100%; and (2) skin severity score (SSS), with ratings from 0 to 45, obtained by measuring skin thickness in 15 areas, using a scale from 0 to 3. A rating of 0 was attributed to normal skin, and 3 in case of severe thickening; intermediate skin involvement was rated as 1 or 2. Patients with an ESS of 0–33%, 34–66% or 67–100% were considered as having mild, moderate or severe skin involvement respectively. Also for SSS, patients were divided into three groups: 0–15 (mild), 16–30 (moderate) and 31–45 (severe). Skin biopsies were obtained from representative areas in 13 patients prior to ECP.

Pulmonary function tests were performed before starting ECP in cGVHD patients. The obstructive ventilatory defect was measured as reduction of maximal airflow with respect to normal. In particular, the severity of the obstructive abnormality was graded as: ‘mild’, ‘moderate’ and ‘severe’, according to the rules established by the American Thoracic Society (1991). Tests were repeated after 6 months and at the end of treatment. Gastrointestinal infectious disease was ruled out by routine diagnostic tests, i.e. cultures for bacterial, fungal (Candida) and viral (CMV, adenovirus) agents, and by searching Clostridium difficile toxins and antigens for rotavirus and adenoviruses. Gastrointestinal biopsy was performed before ECP in 10 patients. Liver involvement was assessed by liver function tests before starting ECP and before each subsequent cycle of ECP. Liver biopsy was performed before starting ECP in seven patients.

Eligibility criteria

Eligibility criteria for ECP treatment were: (1) confirmed diagnosis of either steroid-resistant aGVHD or cGVHD failing to respond to at least two lines of treatment (specifically, steroid resistance was defined as lack of stable clinical improvement after treatment with prednisolone at a dosage of 2–5 mg/kg/d for at least 7 d); (2) complete haematological remission and full donor chimaerism; (3) white blood cell (WBC) count > 1 × 109/l; (4) body weight greater than 10 kg; and (5) no concomitant treatment with either ATG or monoclonal antibodies causing lymphocyte lysis.

aGVHD

Thirty-three patients were given ECP for aGVHD at a median time of 45 d (range 13–98 d) after HSCT. All these patients had skin involvement (grade IV, n = 8; grade III, n = 15; grade II, n = 9; grade I, n = 1). Twenty patients had gastrointestinal GVHD (grade IV, n = 5; grade III, n = 2; grade II, n = 8; grade I, n = 5). Fifteen patients had liver involvement (grade IV, n = 1; grade III, n = 3; grade II, n = 8; grade I, n = 3). The overall clinical stage of aGVHD was: grade I, n = 2; grade II, n = 11; grade III, n = 13; and grade IV, n = 7. The two patients with overall grade I GVHD were treated with ECP because of side-effects due to immunosuppressive drugs. In this group of patients, the median Lansky/Karnofsky performance score at the start of ECP was 60% (range 30–90%).

cGVHD

Forty-four patients had cGVHD, which was extensive in 38 patients and limited in six patients. Progressive cGVHD evolved directly from active aGVHD in 26 patients, whereas the quiescent form of the disease was diagnosed in 13 children. The remaining five patients were diagnosed with de novo cGVHD. In detail, based on the ESS, five patients had mild, nine had moderate and 22 had severe skin involvement and, according to the SSS, 10 patients had mild, 15 had moderate and 11 had severe skin involvement. Joint contracture was found in 14 children (mild, n = 4; moderate, n = 7; severe, n = 3). Lichen planus and sicca syndrome affected 26 and 20 children respectively. A confirmatory skin biopsy was available for 11 patients. Gastrointestinal cGVHD was diagnosed in 21 children (mild, n = 8; moderate, n = 6; severe, n = 7). Twenty patients had liver involvement. Gut biopsy was performed in seven patients and liver biopsy in five.

On the basis of pulmonary function tests, 14 of the 44 patients were diagnosed as having pulmonary involvement (mild, n = 2; moderate, n = 5; severe, n = 7).

In the group of patients with cGVHD, the median Lansky/Karnofsky play performance score at the start of treatment was 60% (range 30–90%). The median interval between diagnosis of cGVHD and start of ECP was 8·9 months (range 0·4–109 months). All but one of the patients had failed at least two lines of immunosuppressive treatment, as detailed in Table I.

Photopheresis procedure

At the Padova unit, ECP was performed using the UVAR® Photopheresis Instrument (Therakos, Exton, PA, USA). In adult patients, the procedure requires that 240 ml of buffy coat and 300 ml of plasma be collected and added to 200 ml of normal saline. However, given the lower body weight of our patients, lower volumes of buffy coat and plasma were chosen, as reported elsewhere (Rossetti et al, 1996; Dall'Amico et al, 1997; Dall'Amico & Zacchello, 1998). After the addition of 100 μg of 8-methoxypsoralen (8-MOP; Gerot, Vienna, Austria) in aqueous solution, the buffy coat and plasma were passed as a thin film through a disposable plastic device, exposed to UVA light (2 Joule/cm2) for 90 min and then returned to the patient.

In children with a body weight lower than 15 kg, the haemoglobin level was maintained above 12 g/dl. To this purpose, a filtered and irradiated red blood cell unit was available for transfusion during the procedure.

At Pavia, Genova and Monza, ECP was performed as follows. A Spectra Cobe cell separator (Lakewood, CO, USA; version 4·6 and 6·0) processing two volumes of the patient's blood was used. The maximum final volume of mononuclear cells (MNC) to be collected was set at 150 ml, with a haematocrit lower than 5%. The MNC collected were adjusted to a constant volume of 300 ml by adding normal saline and 3 ml of 8-MOP to a final concentration of 200 ng/ml. The buffy coat diluted in this way was transferred to a special UVA-permeable bag (Maco Pharma, Tourcoing, France), and UVA irradiation was performed with a UVAMATIC irradiator (Vilber Lourmat, Marne-la-Vallée, France) at 2 Joule/cm2. The 8-MOP photo-activated product was then re-infused into the patient within 30 min.

The average duration of the procedure was 180–240 min for both techniques. During ECP, patients were monitored for blood pressure, heart rate and body temperature. Full blood cell count, liver and kidney function tests, and coagulation parameters were obtained before each procedure. All adverse effects observed during the procedures were recorded. No patient required sedation during the procedure.

Vascular access

A pre-existing 7–9 French Hickman double-lumen central venous catheter was used in 41 patients, while a 7–9 French Arrows double-lumen central venous catheter was placed ad hoc in nine patients, when a minimum flow rate of 10 ml/min was not attainable through a peripheral vascular access.

Treatment protocol

As a rule, patients were treated with ECP on 2 consecutive days at 1-week intervals for the first month, every 2 weeks during the second and third month, and then at monthly intervals for at least 3 further months. In nine patients with grade II–IV aGVHD, ECP was initially performed three times a week on alternate days until clinical improvement was seen. Progressive tapering and discontinuation of ECP was decided upon evaluation of individual response. Any concomitant immunosuppressive therapy was initially maintained, then modified or discontinued according to the clinical response.

Efficacy parameters

Criteria for defining response to ECP were as previously reported (Greinix et al, 1998, 2000):

aGVHD.  All patients that had enrolled for ECP before d + 100 were included in this group irrespective of the fact that they subsequently developed cGVHD. Responders were defined as having an overall clinical grade 0–I after ECP (sixth month). Patients with a greater than 50% response in terms of organ involvement were considered as partial responders, irrespective of their baseline overall clinical stage. Less than 50% response in organ involvement was considered as stable disease, irrespective of any change in immunosuppressive therapy. Any worsening of organ involvement, as well as the appearance of new signs or symptoms of GVHD, was defined as progressive disease. Patients with stable or progressive disease were considered non-responders.

cGVHD.  This group included all patients enrolled for ECP after d + 100. Complete organ response was defined as complete regression of skin, liver, gastrointestinal, lung, oral, joint or eye manifestations. Patients with greater than 50% response in terms of organ involvement were considered as partial responders, irrespective of their original overall clinical stage. An evaluated response of less than 50% in organ involvement was judged as stable disease, irrespective of any changes in immunosuppressive therapy; worsening of organ involvement, as well as onset of new GVHD-related signs or symptoms, was defined as progressive disease. Patients with either stable or progressive disease were considered non-responders.

Statistical analysis

Data were collected on specifically designed forms and were analysed as of 31 December 2001. Overall survival was calculated by the Kaplan–Meier product limit method, the only event considered being the patient's death. Results are expressed as probability (%) and 95% confidence intervals (95% CI). Differences between Kaplan–Meier probabilities were evaluated using the log-rank test. The following patient or graft characteristics were analysed for their potential prognostic value on outcome: patient characteristics (age, sex, diagnosis), transplant-related factors (source of stem cells, type of conditioning regimen, type of donor), and median interval between diagnosis of GVHD and start of ECP. For these prognostic analyses, continuous variables were categorized as follows: each variable was first divided into five categories at approximately the 20th, 40th, 60th and 80th percentiles. If the relative event rates (ratio of the observed number of events to the expected number of events in a category, assuming no variation across categories) in two or more adjacent categories (and the mean times to event) were not substantially different, these categories were grouped. If no clear pattern was observed for the primary outcome, the median was taken as the cut-off point. The sas software package (SAS Institute, Cary, NC, USA) was used for the analysis.

Results

Clinical evaluation of the patients was registered at baseline, after 1, 2, 3 and 6 months, and when ECP was discontinued.

Patients with aGVHD

The median duration of treatment was 74 d (range 8–467 d) for a median number of eight cycles (range 2–20 cycles). At the end of treatment, 18/33 (54%) patients had an overall clinical stage 0–I (i.e. complete response), 7/33 (21%) had a partial response and 8/33 (24·3%) were non-responders (Table II). Among the 18 patients showing complete response, two patients had grade I, six patients had grade II, eight patients had grade III and two patients had grade IV aGVHD. Seven out of these 18 were classified as having an overall clinical stage I (before ECP, three had grade II disease, three had grade III and one had grade IV). Response to treatment, evaluated according to organ involvement, is summarized in Table III. A complete response of aGVHD manifestations of skin, gut and liver was present in 76%, 75% and 60% of patients, respectively, a median 6·4 months (range 1–42 months) after discontinuing ECP. Maximal response to ECP was observed after 8 weeks of treatment (16 procedures). The 5-year overall survival of patients responding to ECP was significantly better than that of non-responders, namely 69% (95% CI 50–81) vs 12% (95% CI 0–34) (P = 0·001) (Fig 1). At the time of writing, 19 patients were alive and 15 of them (79%) had no sign of GVHD. As a result of ECP, it was possible to discontinue immunosuppressive therapy in eight patients (42%) and reduce it in seven patients (36%). The median Lansky/Karnofsky performance score improved significantly from 60% before ECP to 100% (range 80–100%) after completing the treatment. Fourteen patients died while being treated with ECP, as a result of complications of GVHD (n = 10), interstitial pneumonia (n = 2), or relapsing primary disease (n = 2).

Table II.  The outcome of patients in the aGVHD group, according to the initial overall clinical stage.
 CompletePartialNo response
  • *

    The two patients with grade I overall clinical stage showed a complete resolution of GVHD.

  • The number of patients allocated to grade I, II, III, IV GVHD, respectively, before starting ECP is indicated. The response was defined ‘complete’ when the overall clinical stage at the end of ECP was 0–I.

Grade I (2/33) 2/2*  
Grade II (11/33) 6/11 3/11 2/11
Grade III (13/33) 8/13 3/13 2/13
Grade IV (7/33) 2/7 1/7 4/7
54%21%24%
Table III.  The outcome of patients with aGVHD according to organ involvement and grade of aGVHD before ECP.
 CompletePartialNo response
  1. *The patients with grade I in the single organ showed a complete resolution of any sign .

  2. The number of patients allocated to grade I, II, III and IV, respectively, before starting ECP is indicated. ‘Complete response’ was defined when the grade of GVHD in the considered organ was 0–I at the end of ECP. ‘Partial response’ was defined when the response was greater than 50% in organ involvement at the end of ECP, irrespective of basal overall clinical stage. ‘No response’ was defined when either progressive or stable disease, or a response lower than 50% in organ involvement, was observed at the end of ECP.

Skin
 Grade IV (8/33) 4/81/8 3/8
 Grade III (15/33)12/151/15 2/15
 Grade II (9/33) 8/9  1/9
 Grade I (1/33) 1/1*  
76%6%18%
Gut
 Grade IV (5/20) 3/5  2/5
 Grade III (2/20)   2/2
 Grade II (8/20) 8/8  
 Grade I (5/20) 4/5*  1/5
75% 25%
Liver
 Grade IV (1/15)   1/1
 Grade III (3/15) 2/3  1/3
 Grade II (8/15) 4/8  4/8
 Grade I (3/15) 3/3*  
60% 40%
Figure 1.

Acute GVHD: survival according to the response to ECP. The 5-year overall survival of patients responding to ECP was significantly better than that of non-responders, being 69·5% (95% CI 50·2–80·9) and 12·5% (95% CI 0–34·5) respectively (P = 0·001).

Patients with cGVHD

Out of 44 children, 34 (77%) survived and 59% experienced a significant improvement after ECP. The overall response, evaluated according to organ involvement, is detailed in Tables IV and V. The median improvement in ESS and SSS scores was 26% and 8 points respectively. At the end of ECP treatment, 15 (44%) and 10 (29%) of the 34 surviving patients had a complete and partial response respectively. As a result of treatment with ECP, it was possible to discontinue immunosuppressive therapy in 15 (44%) and to reduce it in 10 (29%) of these patients respectively. The median Lansky/Karnofsky score improved from 60% to 90% (range 60–100%). As shown in Fig 2, the 5-year overall survival was significantly better (P = 0·04) in patients responding to ECP than in non-responders, namely 96% (95% CI 89–100) vs 58% (95% CI 34–82). A clinical response was more frequently (64%) observed in patients who started ECP earlier (before the median time of 8·9 months) after cGVHD diagnosis. Age, sex, diagnosis, type of donor and source of stem cells did not influence the outcome in univariate analysis. Ten patients died while being treated with ECP: six of GVHD, two of CMV pneumonia and two of Gram-negative sepsis.

Table IV.  The outcome of patients with cGVHD of the skin.
 ImprovedStableWorsened
  1. According to the area of skin involvement (ESS), patients were divided in three groups: 0–33%‘mild’, 34–66%‘moderate’, 67–100%‘severe’. According to the skin severity score (SSS), patients were divided in three groups: 0–15 ‘mild’, 16–30 ‘moderate’, 31–45 ‘severe’. The patients were considered improved when differences in skin scores greater than 15% from baseline were observed, stable when skin involvement was unchanged and worsened when disease progressed.

Skin
Mild
 ESS 5/36 3/51/51/5
 SSS 10/36 7/103/100
Moderate
 ESS 9/36 6/363/361/36
 SSS 15/36 8/363/364/36
Severe
 ESS 22/3611/367/364/36
 SSS 11/36 5/364/362/36
Table V.  The outcome of patients with chronic GVHD according to organ involvement.
 CompletePartialNo response
  1. In the left column is indicated the number of patients subdivided according to basal organ involvement. ′Complete response′is defined when GVHD disappeared in the single organ. ‘Partial response’ is defined when the response was greater than 50% in organ involvement at the end of ECP. ‘No response’ is defined when either progressive or stable disease or response lower than 50% in organ involvement were observed at the end of ECP.

Gut
 Mild (8/21) 4/8  4/8
 Moderate (6/21) 2/6 2/6 2/6
 Severe (7/21) 1/7 1/7 5/7
33·3%14·3%52·4%
Liver
 Mild (3/20) 2/3  1/3
 Moderate (5/20) 4/5  1/5
 Severe (12/20) 2/12 4/12 6/12
40%20%40%
Joints
 Mild (4/14) 3/4  1/4
 Moderate (7/14) 1/7 3/7 3/7
 Severe (3/14) 1/3  2/3
36%21%43%
Lung
 Mild (2/14)   2/2
 Moderate (5/14) 3/5  2/5
 Severe (7/14) 1/7 2/7 4/7
30%14%57%
Figure 2.

Chronic GVHD: survival according to the response to ECP. The 5-year overall survival was significantly better in patients responding to ECP than in non-responders, being 96·1% (95% CI 88·7–100) and 58·4% (95% CI 34·5–82·4) respectively (P = 0·04).

Side-effects

Side-effects observed during ECP were generally mild, but were more frequent in children with aGVHD and in those with low body weight. ECP caused mild hypotension in 21 patients and abdominal pain in eight patients; these adverse effects did not make it necessary to suspend the procedure. A transient reduction in platelet and/or WBC count was observed in eight patients. A greater than 2 g/dl drop in haemoglobin was reported in 20/33 children treated for aGVHD. All patients with aGVHD required filtered, irradiated RBC transfusion and platelet support during the first eight procedures. One patient with grade IV aGVHD on high-dosage steroid therapy (5 mg/kg/d) experienced acute gastrointestinal bleeding after the second course of ECP: gastrointestinal endoscopy showed multiple ulcers in the stomach. Another thrombocytopenic patient with extensive cGVHD of gut, skin and liver had an acute pulmonary haemorrhage after the sixth course of ECP. It was not possible to exclude a cause/effect relationship between the pulmonary bleeding and either thrombocytopenia or the use of anticoagulants during ECP in these patients. Both patients recovered. Only two central line infections were observed.

Discussion

Several studies have demonstrated that T lymphocytes can be functionally inactivated in vitro by treatment with 8-MOP and UVA (Bredberg & Forsgren, 1984; Deeg et al, 1992; Kapoor et al, 1992; Barr, 1996; Yoo et al, 1996; Aringer et al, 1997; Bladon & Taylor, 1999; Berger et al, 2001). This has been confirmed in an experimental bone marrow transplant (BMT) murine model, where the photo-inactivation with 8-MOP and UVA of donor cells prior to infusion prevented the onset of GVHD (Ullrich, 1991). T-cell function impairment has also been reported in patients undergoing psoralen plus ultraviolet A (PUVA) therapy, an effective treatment for various cutaneous diseases, including skin manifestations of cGVHD (Kapoor et al, 1992). In the light of this evidence, and of previous reports showing the efficacy of ECP in several T cell-mediated diseases (Edelson et al, 1987; Rook et al, 1990, 1999; Costanzo-Nordin et al, 1992; Knobler et al, 1992; Dall'Amico et al, 1995; Barr et al, 1998), it was suggested that ECP could downregulate the alloreactivity of T cells, which play a pivotal role in the pathogenesis of GVHD (Lambert et al, 1989; Berger et al, 1990; Girardi et al, 1995; Ware et al, 1995). Several authors have reported the outcome in adult GVHD patients treated with ECP (Besnier et al, 1997; Gerber et al, 1997; Greinix et al, 1998; Child et al, 1999; Dippel et al, 1999). Most patients enrolled in those studies had extensive cGVHD that had failed to respond to different combinations of conventional immunosuppressive drugs. This may explain the discrepancy in response rates (ranging from 25% to 80%) reported in these studies. A limited number of patients with aGVHD have also been treated with ECP, with a clinical response rate lower than in patients with cGVHD (Richter et al, 1997; Greinix et al, 1998, 2000). Data on paediatric patients are mainly anecdotal and the few clinical reports available include a small number of patients (Rossetti et al, 1995; Konstantinow et al, 1996; Dall'Amico et al, 1997; Dall'Amico & Zacchello, 1998; Cesaro et al, 1999; Salvaneschi et al, 2001). In this study, we analysed the safety and efficacy of ECP in the largest paediatric cohort with resistant cGVHD or aGVHD reported to date. Although patients with cGVHD were an unfavourably selected population (77% with extensive cGVHD; previous failure of ≥ 2 lines of immunosuppressive therapy), we obtained a significant improvement in all manifestations among the 44 patients treated. Moreover, 34 out of the 44 patients survived, with 15 complete responders and 10 partial responders. Other authors have reported that extensive cGVHD lung or liver involvement has been associated with a poor outcome (Ferrara & Deeg, 1991; Sullivan et al, 1991; Blazar et al, 1997). Our results confirmed recent observation of a better prognosis in cGHVD patients when treated early with ECP (Greinix et al, 1998): the response rate was higher in the patients who started ECP earlier after cGVHD diagnosis. Of note, we also documented a benefit in respiratory function tests. One issue that requires more investigation is the optimal schedule and duration of ECP in cGVHD. Improvement is often slow. Prolonged patient-tailored treatment is probably needed. Although PUVA and ECP use a combination of psoralen and UVA light, their efficacy is different: PUVA was found to be of benefit in skin lesions but not in visceral lesions of cGVHD (Kapoor et al, 1992). Another point addressed by our study is the feasibility of ECP in the setting of paediatric aGVHD. In this context, factors such as low body weight, transfusion dependence and incomplete immunological recovery may render this therapy difficult. Nevertheless, ECP was performed in children with a body weight as low as 10 kg, without significant side-effects. As reported earlier (Cesaro et al, 1999; Greinix et al, 2000), maximal response to ECP was observed after 2 months of treatment. In contrast to that reported by Greinix et al (2000), who found gastrointestinal involvement to be a predictor of poor outcome, we obtained a similar response rate for skin, liver and gut manifestations, with a higher percentage of complete response in grade II–III aGVHD. Forty per cent of the children died while on ECP treatment, as a result of aGVHD or other related complications. Most of these patients were children with end-stage disease in whom ECP was a last resort, particularly during the early years of the study. The possibility of either reducing or discontinuing concomitant immunosuppressive therapies, and steroids in particular, in GVHD patients represent a major advantage for preventing the long-term sequelae of survivors (Bradbuty et al, 1994; Zecca & Locatelli, 2000). The results of this retrospective analysis, together with previously published data on second-line therapies for steroid-resistant moderate-to-severe aGVHD (Deeg et al, 1985; Deeg & Hensley-Downey, 1990; Martin et al, 1991; Anasetti et al, 1994; Hings et al, 1994; Hebart et al, 1995; Van Lint et al, 1998; Pavletic et al, 1999; Przepiorka et al, 2000; Zecca & Locatelli, 2000; Khoury et al, 2001), suggest that ECP can be considered a good option for children not responding to steroid therapy. The mechanism behind the positive effect of ECP in patients with GVHD is still not clear. Psoralens are hydrophobic compounds, which intercalate with DNA base pairs after UVA irradiation, forming photo-adducts with pyrimidine bases, as well as with amino acids and fatty acids. The ability of the cells to repair DNA photo-adducts is inhibited completely when the combined doses of 8-MOP (ng/ml) and UVA (J/cm2), used for ECP, are 50 or more (Edelson, 1989). As only 5–10 × 109 peripheral leucocytes are exposed by 8-MOP and UVA during ECP, the efficacy of the treatment cannot be attributed to the mere inactivation of these cells. It is reasonable to hypothesize that an immunomodulatory effect on alloreactive T-cell populations, exposed to 8-MOP plus UVA, is triggered by the treatment (Lambert et al, 1989; Deeg et al, 1992; Girardi et al, 1995; Ware et al, 1995; Barr, 1996; Yoo et al, 1996; Aringer et al, 1997; Bladon & Taylor, 1999; Berger et al, 2001).

In conclusion, our results suggest that ECP is a useful therapy for children with GVHD that is resistant to conventional treatment. Considering that our study may have the intrinsic limit of all retrospective evaluations where patients are not prospectively allocated to treatment, the role of ECP in the treatment of GVHD needs to be confirmed by larger prospective randomized studies.

Acknowledgments

This work has been partly supported by grants from AIRC (Associazione Italiana Ricerca sul Cancro), CNR (Consiglio Nazionale delle Ricerche), MURST (Ministero dell'Università e della Ricerca Scientifica e Tecnologica), IRCCS (Istituto di Ricovero e Cura a Carattere Scientifico) Policlinico San Matteo to F.L., and Fondazione Città della Speranza to C.M.

We would also like to thank Sandra Volpato, PhD, for statistical advice, Stefania Varotto, MD, data manager, the nursing staff, and Nella Augusta Greggio, MD, who introduced us to ECP.

Roberto Dall'Amico is grateful to IMX-38 X-Plosion for continuous support.

Appendix

List of Institutions which enrolled patients in the study:

  •  Brescia, Clinica Pediatrica (Professor L. Notarangelo; F. Porta, MD; Professor A. G. Ugazio);

  •  Genova, Istituto ‘G. Gaslini’ IRCCS (G. Dini, MD; E. Lanino, MD);

  •  Monza, Clinica Pediatrica (Professor G. Masera; C. Uderzo, MD);

  •  Pavia, OncoEmatologia Pediatrica (F. Locatelli, MD; G. Giorgiani, MD; M. Zecca, MD);

  •  Padova, OncoEmatologia Pediatrica (Professor L. Zanesco; C. Messina, MD);

  •  Pesaro, Divisione di Ematologia (Professor Lucarelli);

  •  Pescara, Divisione di Ematologia (P. Di Bartolomeo, MD);

  •  Torino, Clinica Pediatrica (Professor E. Madon; F. Fagioli, MD);

  •  Trieste, Istituto per l'Infanzia Burlo Garofalo (Professor P. Tamaro; Andolina, MD; M. Rabusin, MD).

Ancillary