Stem cell transplantation for indolent lymphoma and chronic lymphocytic leukemia

Abstract

The indolent lymphomas, including chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL) remain incurable with standard therapy. Autologous hematopoietic stem cell transplantation (HSCT_[JHA1]) is feasible and has low treatment related mortality in follicular lymphoma, but there are questions relating to optimal timing of the procedure, conditioning regimen and late effects. Myeloablative allogeneic HSCT is associated with high treatment related morbidity and mortality, few late relapses, but is applicable to only a small number of patients. The major focus of HSCT in these lymphomas has been with reduced intensity conditioning (RIC) allogeneic HSCT, which is applicable to the age distribution of these diseases and which exploit the graft versus lymphoma effect in these diseases. Steps to further decrease the morbidity and mortality of the RIC HSCT and in particular to reduce the incidence of chronic extensive graft versus host disease while maintaining tumor control remain the major focus. Many potential treatments are available for indolent lymphomas and CLL, and appropriate patient selection and the timing of HSCT remain controversial. The use of HSCT must always be weighed against the risk of the underlying disease, particularly in a setting where improvements in treatment are leading to improved outcome.

Stem Cell Transplantation for Indolent Lymphoma and CLL

Hematopoietic stem cell transplantation (HSCT) has become the treatment of choice for patients with relapsed aggressive non-Hodgkin lymphoma (NHL) and Hodgkin lymphoma (HL). To date, most of these patients have been treated with autologous stem cells, currently using peripheral blood stem cells mobilized by chemotherapy and/or recombinant growth factors. The role of HSCT in the management of patients with indolent lymphoma remains more controversial although increasing numbers of patients with advanced stage follicular lymphoma, mantle cell lymphoma and chronic lymphocytic leukemia are now undergoing HSCT. There is increasing concern regarding a continuous risk of relapse of disease after autologous HSCT, especially if done late in high risk patients and in the long term toxicity associated with this approach, especially the development of secondary myelodysplastic syndrome and other cancers. This has led to renewed interest in the role of allogeneic HSCT for patients with indolent lymphoma; although major problems remain with the high treatment related mortality (TRM) associated with myeloablative allogeneic HSCT as well as graft versus host disease (GVHD). The advantage of allogeneic HSCT is the potential to exploit the graft versus lymphoma (GVL) effect. Many studies are underway exploring the use of reduced intensity conditioning (RIC) regimens, steps to control disease progression and manipulate donor cells to maximize T cell responsiveness against lymphoma and minimize GVHD.

Patient selection

Unlike aggressive lymphomas, the use of high-dose chemotherapy with autologous HSCT in the treatment of indolent lymphomas has not yet been fully established. The rationale for considering transplantation is that the disease is incurable using standard approaches and that younger patients with these “indolent lymphomas” will die of their disease and that promising results have been observed in a number of phase II studies.(1-3) Detection of MRD has been a useful surrogate marker for tracking long-term PFS in patients, in examining tumor contamination of autologous stem cells and in monitoring serial samples after transplantation.(3-7) A major concern relates to the risk of development of development of secondary myelodysplasia/acute myeloid leukemia, especially with the use of high dose_[JHA2] total body irradiation (TBI) containing conditioning regimens (8). The role of high dose therapy and autologous HSCT in FL patients during first remission has been explored in phase II trials (9-10), and in four phase III randomized trials (11-14). Three hundred and seven previously untreated patients, up to 60 years of age who responded to induction chemotherapy were randomized to autologous HSCT or IFN-alpha maintenance. (11) Among 240 evaluable patients, the 5-year PFS was 64.7% for autologous HSCT, and 33.3% in the IFN-alpha arm (P < .0001). Acute toxicity was higher in the autologous HSCT group, but early mortality was less than 2.5% in both study arms. One hundred seventy two newly diagnosed advanced FL patients were randomized to combination chemotherapy and IFN-alpha or to high-dose therapy followed by purged autologous HSCT (12). Patients treated with autologous HSCT had a higher response rate than patients who received chemotherapy and IFN-alpha (81% versus 69%, p=0.045) and a longer median progression-free survival (PFS) (not reached versus 45 months), but this did not translate into a better overall survival (OS) due to an excess of secondary malignancies after transplantation. Four hundred and one previously untreated advanced stage FL patients were randomized to receive combination chemotherapy and IFN-alpha compared with four courses of CHOP chemotherapy followed by high dose therapy using TBI and autologous HSCT. Intent-to-treat analysis after a median follow-up of 7.5 years showed no difference between the two arms for OS (p = 0.53) or PFS (p = 0.11). Long-term follow-up demonstrated no statistically significant benefit in favor of first-line ASCT in patients with FL. Only one study used rituximab containing chemoimmunotherapy.(14) Patients relapsing after CHOP-R underwent salvage therapy with autologous HSCT. No OS advantage was seen for high dose therapy. In view of these results, autologous HSCT should be used in first remission only in the setting of clinical trials.

The European Bone Marrow Transplant Registry (EBMTR) sponsored CUP study (conventional chemotherapy, unpurged, purged autograft) is the only prospective randomized trial to assess the role of autologous HSCT in patients with relapsed FL.(15) The results of the study suggest a PFS and OS advantage of ASCT over conventional chemotherapy, with 4 year OS of 46% for the chemotherapy arm, versus 71% for the unpurged and 77% for the purged ASCT arms. The study was closed early because of slow accrual with 140 of the planned 250 patients accrued and only 89 randomized. In CLL/SLL, the use of HSDT and ASCT was not associated with improved outcome in patients transplanted in first remission compared to those transplanted later in their disease course.(16)

Autologous HSCT is not a suitable treatment option for most patients with CLL. The disease usually follows an indolent course; many patients never require any therapy and most patients are too elderly to undergo this procedure. In addition, treatment with fludarabine or other purine analogues makes collection of autologous PBSC difficult and tumor cell contamination of the bone marrow is high. High-risk patients can be identified using a number of clinical and biological features and such younger patients are suitable candidates for enrolment in clinical trials evaluating the role of HSCT in CLL. The role of HSCT has been in prospective studies in a number of other hematologic malignancies established, but no studies in CLL published to date have compared the outcome after standard chemotherapy with either autologous or allogeneic HSCT. Using clinical and biological features it is possible to identify patients who are suitable candidates for enrolment in clinical trials evaluating the role of HSCT. The biggest challenges remain the decision of which patients are eligible for consideration of HSCT and when in their disease course HSCT should be offered.

No studies have prospectively compared the role of standard chemotherapy with autologous HSCT in CLL and there is no established role for this approach except in the setting of a clinical trial. A retrospective matched-pair analysis has suggested a survival advantage for autologous HSCT compared with conventional therapy (17). This used a risk-matched comparison between 66 patients who had undergone a uniform high dose therapy and autologous HSCT with a database of 291 patients treated conventionally. Matched data included age, Binet stage, IgV_H mutational status and lymphocyte count and 44 patient pairs matched all four variables. With an overall median follow-up time of 70 and 86 months, survival was significantly longer for the patients who had undergone autologous HSCT compared with conventionally treated patients when calculated from diagnosis (p=.03) or from study entry (p=.006).

A number of phase II studies have reported outcome after autologous HSCT for CLL (18-20). These studies have demonstrated that this approach is feasible in CLL with a transplant transplant-related mortality (TRM) of 1 to 10%, with most toxicity occurring late. Among 115 previously untreated CLL patients prospectively enrolled in a pilot study to assess the feasibility of performing autologous HSCT only 65 (56%) proceeded to transplant.(19) Only one TRM was seen and the CR rate after transplantation was 74% (48 of 65). The 5-year estimated OS was 77.5% and PFS was 51.5%. None of the variables examined at study entry were predictive for OS or PFS, but detectable minimal residual disease (MRD) was highly predictive of disease recurrence. Most studies reported have relatively short follow up and therefore focus only on TRM early post transplant, but late consequences, particularly development of secondary myelodysplasia and acute myeloid leukemia (MDS/AML), are of concern and a high incidence of other solid tumors in 31 (19%) patients (18).

Myeloablative allogeneic HSCT

Allogeneic transplantation was initially pioneered for the treatment of acute and chronic leukemias and patients with indolent lymphoma were seldom offered this approach. There were excellent outcomes following autologous HSCT in phase II studies and in a small randomized clinical trial, a clinical benefit was seen for auto-SCT over conventional chemotherapy in FL (15). Studies of minimal residual disease showed a strong association between lymphomatous involvement of the autologous graft and disease recurrence (21). This provided the initial rationale for myeloablative allogeneic HSCT, which was initially used in patients with extensive marrow involvement or those who had failed previous auto-HSCT. Total body irradiation containing regimens were largely used. An International Bone Marrow Transplantation Registry (IBMTR) registry analysis of 113 patients reported over a 12-year period (1984–1995) by 50 teams worldwide, an average of 2 patients per team.(22) The registry analysis confirmed a very high rate of durable disease control after allogeneic HSCT for indolent lymphoma and only a 16% incidence of disease recurrence. However, TRM was high and adverse outcome and high TRM was seen in patients with refractory disease, decreased performance status or age over 40 years. Very similar results were seen with EBMT registry analysis which also found similar OS and PFS for myeloablative allogeneic HSCT (23). In a case control analysis they reported less relapse after allogeneic compared to autologous HSCT, but survival after allogeneic HSCT appeared inferior due to the increased TRM. Data suggest that the outcome of myeloablative transplant is improving over time, most likely related to more judicious patient selection as well as better supportive care.(24)

In CLL, allogeneic HSCT has significant morbidity and mortality, from regimen-related toxicity, GVHD and infection, but surviving patients may have long term disease control (18,25-27). In registry data, TRM following allogeneic HSCT in CLL patients was unacceptably high at 46%, with mortality from GVHD of 20% (25). Of 25 patients with CLL who underwent allogeneic HSCT at the Fred Hutchinson Cancer Center grades 2-4 acute GVHD was seen in 14 patients and 10 developed clinical extensive chronic GVHD and estimated OS at 5 years was 32% (27). Non-relapse mortality at day 100 was unacceptably high at 57% for patients conditioned with busulfan and cyclophosphamide compared to 17% for patients conditioned with TBI containing regimens. Among 30 patients (20 related donors and 10 unrelated donors) transplanted for CLL between 1989 and 2001 in Vancouver with a median follow-up of 4.3 years, 47% were alive in CR, both estimated OS and DFS at five years were 39%. A strong graft versus leukemia (GVL) effect was noted, with those developing acute or chronic GVHD having near complete protection from relapse.

There are no randomized studies comparing the outcome of autologous versus allogeneic HSCT. In a report of the IBMTR, results are described for 904 patients with FL (28). Among these patients, 176 patients underwent allogeneic HSCT, 131 patients underwent autologous HSCT using purged stem cells and 597 using un-purged autologous stem cells. The treatment related mortality (TRM) in these three groups was 30%, 14% and 8% respectively, disease recurrence in 21%, 43% and 58% and 5 year overall survival was 51%, 62% and 55% respectively. The use of TBI containing regimens was associated with increased TRM but decreased risk of relapse. The use of allogeneic HSCT was associated with increased TRM, but significantly lower risk of disease recurrence consistent with graft versus lymphoma effect and possibly the use of a tumor free stem cell source. Further steps that decrease the TRM after allogeneic HSCT are therefore likely to result in improved outcome in these disease, with resulting improvement in outcome after allogeneic compared with autologous HSCT, although long term follow up will be required to confirm this. In CLL, studies from MD Anderson Cancer Center demonstrate improved outcome after allogeneic compared to autologous HSCT (29) suggesting that allogeneic HSCT can induce durable remission even in patients with refractory disease. At Dana-Farber Cancer Institute, 162 patients with high risk CLL were enrolled in a “biologic randomization” in which 25 patients with an HLA matched sibling donor underwent T cell depleted myeloablative allogeneic HSCT while 137 with no HLA matched sibling donor underwent B cell purged autologous HSCT, with both groups receiving identical conditioning regimen using high dose cyclophosphamide and TBI.(18) The 100-day TRM was 4% after autologous or allogeneic HSCT, but later TRM had a major impact on outcome. At the median follow-up of 6.5 years, PFS was significantly longer following autologous than T cell depleted allogeneic HSCT, but no significant differences were observed in disease recurrence or deaths without recurrence by type of transplant. There was no difference in OS between the two groups and at the median follow up time of 6.5 years OS was 58% after autologous and 55% after allogeneic HSCT.

4. Reduced intensity conditioning HSCT

A major advance in reducing the short-term morbidity and mortality of allogeneic HSCT has been the introduction of non-myeloablative or reduced intensity conditioning (RIC) regimens to allow engraftment of allogeneic stem cells. Most patients to date have been treated on experimental treatment protocols which allowed enrolment of many patients with chemo-refractory end-stage disease.

RIC regimens allow transplantation in older patients, making this approach more applicable to increased numbers of patients. Results from the larger reported studies in indolent lymphoma (24,30-33) are shown in Table 2 and for CLL (34-39) in Table 3. Most patients were heavily pre-treated and many were already refractory to therapy, but despite this the majority demonstrated donor engraftment and there is a high CR rate. The ability of such approaches to eradicate MRD in patients with advanced CLL (40) and the observation of late remissions in patients treated with low doses of chemotherapy provide the strongest direct evidence for a powerful GVL in CLL. The outcome from the Fred Hutchinson Cancer Research Center multi-institutional protocol after RIC allogeneic HSCT for 82 patients with advanced fludarabine refractory CLL using related (n=52) or unrelated donors (n=30) median age 56 (range 42-72) years demonstrated TRM of 23% at 5 years, with significant GVHD remaining a problem. (34) Five-year OS was 50% and DFS was 39%. Although complications were higher in the patients with unrelated donors, there were higher CR and lower relapse rates, suggesting more effective GVL activity with unrelated donors. Forty-six patients underwent RIC transplantation at Dana-Farber Cancer Institute, 67% using unrelated donors.(36) Factors associated with increased risk of relapse include low levels of donor chimerism at day 30, chemo-refractory disease, increased number of previous therapies and adverse cytogenetics.(36)

Table 2.

RIC allogeneic HSCT for FL.

n	Age years (range)	Regimen	Chemo-Refractory (%)	Prior Auto	Donor (related/unrelated)	TRM	Chronic GVHD	Survival	Reference
47	53 (33-68)	Flu/CY/Rit	0	-	96% 4%	15%	60%	OS 85% 5 yr PFS 83%	Khouri et al 2008 (30)
62*	54 (33-66)	(Flu)/tbi	33%	26%	55% 45%	42%	47%	OS 52% 3 yr PFS 43%	Reznavi et al 2008 (31)
88	51 (21-70)	Various	31%	-	100%	28%	62%	OS 62% 3 yr PFS 55%	Hari et al 2008 (24)
73	54 (37-67)	Flu/Bu/ATG Various	26%	-	86% 14%	15%	43%	OS 55% 3 yr PFS 53%	Vigouroux et al (32)
52	46 (27-65)	Various	15%	-	91% 9%	31%	17%	OS 65% 1 yr PFS 54%	Robinson et al 2002 (33)

^*16 had transformed

Table 3.

RIC allogeneic HSCT for CLL.

n	Age years (range)	Prior regimens (range)	Chemo-Refractory (%)	Prior Auto-SCT	Donor (includes mismatch)	TRM	GVHD acute gd 2-4	chronic extensive	Survival	Reference
82	82 (42-72)	4	87%	4	63% related 37% unrelated	25%	55%	49% related 53% unrelated	OS 50% 5 yr PFS 45%	Sorror et al 2008 (34)
77	54 (30-66)	3 (0-8)	33%	10	81% related	18% 12m	34%	58%	OS 72% 2 yr PFS 56%	Dreger et al 2003 (35)
46	53 (35-67)	5 (1-10)	57%	10	33% related 67% unrelated	17%	34%	43%	OS 54% 2 yr PFS 34%	Brown et al 2006 (36)
41	54 (37-67)	3 (1-8)	27%	11	58% related 42% unrelated	5% at 100 d 26%	10% (gd 3-4)	33%* *after DLI	OS 51 2 yr PFS 45%	Delgado et al 2006 (37)
39	57 (34-70)	3 (2-8)	Not stated		90% related 10% unrelated	2% at 100 d	45%	58%	OS 48% 4 yr PFS 44%	Khouri et al 2006 (38)
30	50 (12-63)	3 (0-8)	47%		50% related 50% unrelated	13%	56%	21%	OS 72% 2 yr PFS 67%	Schetelig et al 2003 (39)

The status of PET scan at HSCT does not appear to predict outcome after reduced-intensity allogeneic HSCT (41). This is in contrast with observations in autologous HSCT, where outcome for patients with metabolic evidence of residual lymphoma was significantly worse than in patients with negative PET scans before transplantation. These data suggest that RIC- ^HSCT is able to overcome the unfavorable prognostic effects of a positive PET before transplantation. Any GVL effect is most probable to be effective in the presence of relatively low disease burden and relapse should ideally be detected early to maximize the benefit of RIC-SCT or subsequent DLI. The use of PET surveillance after RIC-SCT was able to detect early relapse that was not apparent on simultaneous CT scans and can be used to direct the use of DLI (41).

No formal assessment of RIC compared to myeloablative allogeneic HSCT has been undertaken in a prospective trial. IBMTR compared traditional myeloablative allogeneic HSCT (n = 120) to RIC-SCT (n = 88) in patients with follicular lymphoma transplanted with HLA-identical siblings between 1997 and 2002. (24) As has been seen in many single center studies, patients receiving RIC tended to be older and have a longer interval from diagnosis to transplant. The median follow-up period after myeloablative allogeneic HSCT was 50 months (4-96 months) compared with 35 months (4-82 months) after RIC-SCT (P < .001). At 3 years, OS was 71% and PFS 67% for the myeloablative allogeneic HSCT group, and for the RIC –SCT OS was 62% and PFS 55%. These differences were not significant. Poor performance score and resistance to chemotherapy were associated with higher TRM, lower OS and PFS. An increased risk of lymphoma progression after RIC-SCT was observed in multivariate analysis, (relative risk = 2.97, P = .04). Of interest, the use of RIC regimens increased from 10% of transplants in 1997 to 80% in 2002, establishing RIC-SCT as a de facto standard for allogeneic HSCT in patients with indolent lymphoma. The differences in patient selection (age, disease bulk, comorbid conditions) for myeloablative versus reduced intensity conditioning make it difficult to draw conclusions regarding the role of high-dose versus reduced intensity conditioning and the effect of GVL immune activity.

For CLL patients, the outcome of 73 patients who had undergone RIC was compared with that of 82 matched patients who had undergone standard myeloablative allogeneic HSCT for CLL from the EBMT registry database during the same time period. Patients undergoing RIC-SCT had significantly reduced TRM, but higher relapse incidence and there was no significant difference in OS or PFS between these two groups (42). Of particular interest is the group of CLL patients with deletion of 17p and loss of p53. A recent report from EBMT of 44 such patients suggests that allogeneic HSCT has the potential to induce long term remission in these very high risk patients (43).

5. Addition of monoclonal antibodies to RIC HSCT

GVHD remains the major concern after RIC HSCT and attempts have been made to utilize monoclonal antibodies to reduce the incidence of GVHD without increasing the subsequent risk of relapse. Excellent results have been obtained using RIC based on a combination of fludarabine, cyclophosphamide with the addition of rituximab at the MD Anderson Cancer Center, an approach designed to maximize GVL by early tapering of immune suppression with use of rituximab and DLI. Among 39 patients treated, median age was 57 (range 34–70) years, median time from diagnosis to transplantation was 4.5 years (38). All patients had recurrent advanced disease, were heavily pretreated with a median of 3 (range 2-8) chemotherapy regimens and all had been previously treated with fludarabine-rituximab-based regimens. At transplant, 34 patients (87%) had active disease, including 9 (23%) with evidence of Richter’s transformation. In this series only four of the donors were unrelated. Fourteen patients required immunomodulation with rituximab and DLI for persistent disease after HSCT. Only one patient died early and among the 38 evaluable patients, 27 (71%) achieved CR, with estimated OS at 4 years was 48% with current PFS was 44%. Acute grade II–IV GVHD was observed in 45%, but chronic extensive GVHD was reduced without concomitant increased risk of relapse.

GVHD can also be decreased using alemtuzumab in the conditioning regimen, but this delays post-SCT immune reconstitution, increases the risk of infective complications and does appear to impair GVL. In 41 consecutive CLL patients treated (24 HLA-matched sibling donors and 17 unrelated volunteer donors, including 4 mismatched) the conditioning regimen alemtuzumab with fludarabine and melphalan had significant anti-tumor effects with 100% of patients with chemosensitive disease and 86% with chemorefractory disease responding (37). The TRM rate was 26%, overall survival 51% and relapse risk 29% at 2 years. GVHD rates were relatively low with acute GVHD occurring in 17 (41%) and chronic GVHD in 13 (33%). The unexpectedly high TRM rate was due to a high incidence of fungal and viral infections_[JHA3].

Exploration of the GVL effect in lymphoma

Following RIC-SCT, the major anti-tumor effect is from the GVL effect. Currently, the targets of the GVL effect and whether GVL can be separated from GVHD is not clear. Donor T cells can recognize host minor histocompatibility antigens that may also be expressed on the tumor cell. If this is the whole basis of GVL, then the therapeutic window between GVL and GVHD will be very narrow. If donor T cells can also recognize tumor-associated antigens in the host theoretically a GVL effect could occur without the major side effect of GVHD and could therefore be exploited in a post transplant setting. Importantly, if this mechanism can be maximized, it should be possible to induce using autologous as well as allogeneic T cells. The outcome in FL is strongly influenced by the immune cell microenvironment (44-46), immune therapy including monoclonal antibodies (47), therapeutic vaccines (48) and rituximab surprisingly enhances immune responses against FL (49). The mechanisms whereby the immune microenvironment affects FL outcome is poorly understood, but tumor cells alter anti-tumor immune responses, including recruitment of pro-tumor macrophages and suppression of cytolytic T-cells (50-51). We have demonstrated that CLL cells induce changes in autologous T cells and that co-culture of CLL cells with healthy allogeneic T cells can induce similar defects (52). The mechanism whereby tumor cells impair cell function appears to be mediated at least in part by actively impairing T-cell actin cytoskeleton, essential for activation and function (53). We have shown that FL-induced T-cell dysfunction in the lymph node occurs by a similar mechanism (54) and that subsets of infiltrating macrophages induce angiogenesis (55). These results define a novel immunosuppressive mechanism induced by FL and CLL cells and identify functional tumor microenvironment biomarkers that should facilitate development of enhanced immunotherapeutic strategies for lymphoma patients. The question remains as to whether remaining tumor cells in the host can also induce an immunosuppressive effect on donor T cells at the time of RIC-SCT and infusion of donor lymphocytes.

We are currently exploring the molecular mechanisms in tumor cell/T cell interactions. Ways that this can be exploited include the use of cord blood expanded immune cells (56-57), expansion of T cells ex vivo, and development of tumor specific T cells for use as DLI. An exciting area is the use of T cells with chimeric antigen receptors. T cells can be genetically modified to recognize putative tumor-associated antigens. In this regard, the receptor tyrosine kinase-like orphan receptor 1 (ROR1) was identified as a highly expressed gene in CLL, but not normal B-cells, suggesting it may serve as a tumor-specific target for therapy. ROR1-specific CAR confer specific recognition of primary B-CLL and MCL including potential tumor initiating progenitor cells (58). T-cell therapies targeting ROR1 may be effective in B-CLL and other ROR1-positive tumors. However, the expression of ROR1 on some normal tissues suggests the potential for toxicity to subsets of normal cells

Conclusions

SCT has a role to play in selected indolent lymphoma and CLL patients, with major focus on the use of RIC allogeneic HSCT HSCT. Although RIC HSCT HSCT appears to result in high response rates and eradication of PCR detectable MRD, the follow up of most clinical trials is too short. However, it does appear that RIC –SCT can cure these diseases. Future approaches to the management of this disease must take into account the balance between the increased morbidity and mortality of HSCT with the curative potential that these approaches potentially offer, in the setting of the improvements in outcome that can now be seen using chemo-immunotherapy. In the absence of any other treatment modalities currently capable of improving outcome in this disease, HSCT should be considered as a treatment approach for younger patients with high-risk lymphoma earlier in the course of the disease, ideally in the setting of well-designed clinical trials assessing the impact of this treatment on outcome in these patients and such trials are currently underway.

Table 1.

Selected results of myeloablative allogeneic HSCT for Indolent Lymphoma

N	Conditioning	Age, y (range)	HLA sib/other donor	Chemo sensitive disease	TRM %	Chronic GVHD	PFS %	OS %	Median F/U (m)	Reference
120	TBI 75% Non-TBI 25%	44 (22–70)	100%	72%	25	46	67	71	50	Hari et al 2008 (24)
231	Various	39 (19–66)	96%/4%	80%	38	–	43	51	60	Peniket et al 2003(23)
113	TBI 84% Non-TBI 16%	38 (15–61)	199%	63%	40	24%	49	49	36	Van Besien 1998 (22)