COMPARACIÓN ENTRE LOS ENSAYOS DE CARGA

In Chapter 3 progenitor thresholds were defined for predicting rapid engraftment after high dose therapy with peripheral blood stem cell support. The general consensus is that a PBSC progenitor dose between 1-5 x10®/kg CD34+ cells or 1-5 xIOVkg GM-CFC is required to ensure rapid engraftment (Demirer & Bensinger, 1995) and that higher progenitor numbers result in more rapid engraftment (Bensinger et al., 1995). Many reports have shown a reduction in mean engraftment times with progressively higher doses but there is still an obligate period of cytopaenia and a steep dose relationship so that gains rapidly diminish above around 1 x10®/kg CD34+ cells (Haas et a!., 1995; Weaver et a!., 1995). Against the need to balance high progenitor yields against over-agressive mobilization and apheresis protocols, particularly in heavily pre-treated patients and to maximise available resources, a pragmatic hazard/risk assessment was adopted in Chapter 3. In this approach the incidence of pre-defined slow (21-28 days) or delayed (>28 days) recovery times to achieve platelet independence or a neutrophil count of 0.5 x1 OVL, were compared with respect to MNC, CD34+ cell and GM-CFC numbers re-infused for 256 patients after high dose therapy. MNC numbers were not predictive in this diverse group of patients although we and others have reported this can be the case using well-standardized mobilization regimens (Scott et a!., 1995a; Watts et al., 1997c). A clear threshold was evident in this series at 1 xIOVkg CD34+ cells re-infused. The 41 patients who received CD34+ cells below this dose included all of the 3 patients with delayed neutrophil recovery and 20 of the 29 patients with delayed platelet recovery including all of those who failed to achieve platelet independence within 70 days. No significant increase in the incidence of delayed platelet recovery was noted among the patients receiving CD34+ cell doses between 1-3.5 xIOVkg (which was 8/102 overall). There was however a significantly reduced risk of

delayed platelet recovery between this group and recovery in those who received a CD34+ cell dose above 3.5 x10®/kg (1/113, P=0.0157) in line with our previous report (Watts et al., 1997c). A reinfused GM-CFC dose below 1 xIOVkg was also highly predictive of poor engraftment and included the three patients with delayed neutrophil recovery and 12 of the 29 patients with delayed platelet recovery. However, it was found in this larger series that 12 patients who experienced delayed platelet recovery were among the 43 patients who received between 1-2 xIOVkg GM-CFC which was not significantly different from those who received less than 1 x1 OVkg GM-CFC. If a GM-CFC threshold were used alone therefore a minimum criterion of 2 xIOVkg would be required. Notably, all of the 12 patients referred to above who had delayed platelet recovery and received 1-2 xIOVkg GM-CFC were also among those patients who received a CD34 dose below 1 x10®/kg. Above 3.5 xIOVkg GM-CFC delayed platelet recovery was rare. We have previously found low CD34+ cell and GM-CFC counts to be closely correlated (Watts et a!., 1997c), and this revised data has increased our previous minimum progenitor thresholds for direct progress to transplant from 1 x10®/kg CD34+ cells and 1 xIOVkg GM-CFC to 1 x10®/kg and 2 xIOVkg for CD34+ cells and GM-CFC respectively. Fortunately when >2 x107kg CD34+ cells are collected 98% of patients will have >2 xIOVkg GM-CFC. We therefore believe that if >2 xIOVkg CD34+ cell are collected it is appropriate to proceed to transplantation without waiting for culture data. It should be noted that haematological engraftment in this thesis refers to short-term engraftment sufficient to ensure independence from platelet support. Haas has shown in a retrospective study of 145 consecutive PBSCT patients that recovery to a platelet count of at least 150 x10®/L within 10 months was more likely in those patients who received threshold levels of 2.5 x10®/kg CD34+ cells or 1.2 xIOVkg GM-CFC. In this study there were 9 patients who received a median of only 0.18 (range 0.02- 0.49 ) xIOVkg GM-CFC who took 1-6 years to achieve normal counts. However none of the 93 patients assessable for long-term engraftment, with a maximum

follow-up of 15 years, developed graft failure (Haas et al., 1995). The clinical implications of slower long-term engraftment are not clear and an audit is in progress for our own patients.

It was noted in Chapter 3 that of the 9 patients with slow neutrophil recovery between 21-28 days that there were 3 in each group receiving CD34+ cell thresholds of <1, 1-3.5 and >3.5 x10®/kg and that none of these patients were among the 81 patients who received early post transplant G-CSF. This is consistent with the beneficial effect of post infusion G-CSF which has been shown in randomised trials (Linch, Klumpp).

In Chapter 4, parameters predicting for progenitor yields were assessed for 101 pre-treated lymphoma patients all mobilized with cyclophosphamide 1.5g/m^ and G-CSF, with the first apheresis performed when the recovery WBC first exceeded 5.0 x 107L. The relationship between the number of progenitor cells collected and patient age, sex, diagnosis, prior radiotherapy and time since last chemotherapy was determined by multivariate analysis.

Prior radiotherapy was found to be predictive for poor mobilization when the progenitor yield was considered as a continous variable in line with other reports (Haas eta!., 1994a; Schwartzberg eta!., 1993). Paradoxically wide field irradiation was not more detremental than local treatment and this may / have been a statistical "quirk" of small numbers. Interesting, however a recent study by Sharp et al showed that cytokine-induced HPC mobilization in mice was inhibited or abolished whether applied to a single limb, upper or lower half of the body. Furthermore, humoural inhibition of mobilization was seen when plasma from an irradiated mouse was injected into a normal mouse prior to cytokine adminstration (Sharp et al., 1998). This implies that impaired HPC mobilization cannot be assumed to only affect the area irradiated.

The prediction of poor mobilization to threshold levels of 1 or 2 x107kg CD34+ cells at first apheresis against a dichotomous variable of <6 or >6 cycles of chemotherapy in Chapter 3 did not achieve significance and only became

suggestive where the number of chemotherapy cycles were considered as a continuous variable (P=0.06 for both 1 and 2 x107kg CD34+ cell thresholds). This lack of association is very similar to the findings in some studies (Ketterer

e ta i, 1998; Kotasek etal., 1992), but not others (Drake et a i, 1997; Haas et ai,

1994a; Schwartzberg e ta i, 1993). In the Drake study a haematological toxicity scoring system for prior chemotherapy agents was devised which showed significantly poorer mobilization between those patients with heavy pre­ treatment and high scores compared to those with low scores. This was not completely predictive however and could not accurately determine which individuals would achieve a given progenitor threshold. This is perhaps not surprising in view of the wide baseline of progenitor mobilization among normal individuals (Roberts et a i, 1995) and the identification of "poor mobilizers" at present therefore relies on failure to achieve a minimal progenitor yield following a well-standardized protocol.

In Chapter 5 the extent of inter-individual progenitor cell mobilization was assessed in 20 healthy male volunteers who were part of a cross-over study to compare the mobilization efficacy of two types of G-CSF, filgrastim (non­ glycosylated) and lenograstim (glycosylated). The average peak levels of GM- CFC mobilized with the glycosylated G-CSF was 28% higher than the non­ glycosylated form but the intra-individual variation between each mobilization was much less than between individuals. There was more than a 10-fold variation between the best and worst mobilizers in agreement with other studies (Hoglund et ai, 1996; Jones et ai, 1996; Roberts et al., 1995; Watts et a i, 1997a). In contrast, even accounting for the difference in G-CSF type administered, the intra-individual progenitor numbers mobilized were relatively similar, in accord with a remobilization study reported by Anderlini (Anderlini et a i, 1997). Notably the worst mobilizer at first G-CSF exposure in Chapter 5 was also the worst on the second occasion. The reason for such wide inter­ individual variabilty in mobilization is unknown, but this study showed it was

unlikely to be due to variations in metabolism that could modulate serum G- CSF concentrations.

Some studies have attempted to predict individuals who will mobilise HPC poorly on the assumption that the incremental rise in circulating progenitors will be fairly constant and can be predicted by measuring baseline or "steady state" circulating GD34+ cell numbers. In a study reported by Fruehauf et al for example, pre-mobilization circulating CD34+ cells were measured in 15 pre­ treated cancer patients and an individual estimate of subsequent yield calculated with a 95% probability of success. A recent study by Brown et ai,

also used baseline pre-cytokine CD344- cell counts was used in a similar way to predict the subsequent progenitor yield in 47 normal donors and is detailed in Chapter 5, p104 (Brown etai., 1997). A major disadvantage of this approach of poor mobilizer identification is the requirement for exceptionally precise flow cytometry. The steady-state blood CD34 numbers for those patients who would require more than one apheresis to achieve the minimum progenitor yield of 2.5 x10®/kg in the Fruehauf report ranged from 1.0-0.2 x10®/L (Fruehauf et a/., 1995). Assuming a WBC of 10 x1 0 7 L for example, this equates to a CD34+ cell concentration of 0.01-0.002%. In the Brown study the baseline counts of healthy donors ranged from 0.01-0.15% (mean 0.05%) which gave absolute counts of 0.8-11.4 x10®/L (mean 4.3 x107L) (Brown et a/., 1997). Many clinical laboratories would be likely to find this level of precision challenging judging by the wide variability in CD34-h cell measurement at much higher levels seen in inter-laboratory quality assurance exercises (Barnett et a/., 1998; Chin-Yee et

a/., 1997; Gratama etai., 1997; Gratama etai., 1998).

Despite the demonstration of a high degree of correlation between the immediate pre-apheresis blood CD34+ cell count of G-CSF mobilized normal donors and baseline CD34+ cell counts or total G-CSF received in the Brown study, these two factors accounted for less than half of the variability in CD34+ cells mobilized (Brown e ta i., 1997), suggesting other important reasons for mobilization variability.

It is possible that since bone marrow stores of stem cells would be expected to exceed a normal lifespan there may be little evolutionary pressure to restrict random genetic variation in stem cell frequency above the minimum level required. Some evidence for this notion has been demonstrated in mice where the chromosomal location of "stem cell frequency genes" has been identified. In separate studies reported by Muller-Sieberg and Roberts, a narrow range of LTCIC numbers was seen within individual mouse strains but at least a 10-fold variation between strains (Muller-Sieburg & Riblet, 1996; Roberts et al., 1997). In a subsequent report, Muller-Sieberg described two mice strains with high marrow LTCIC frequency in one instance and low in the other. Both types were mobilized with G-CSF and showed the mobilization of blood LTCIC in proportionally high or low numbers.

In Chapter 6 the value of HPC support provided by the collection of autologous bone marrow was assessed for those patients who failed to collect a minimum yield of peripheral blood stem cells. The lowest level of PBSC CD34+ cells to ensure engraftment is thought to be 1 xIOVkg (Bensinger et al., 1994; Watts at al., 1997c) and there was a failure to harvest this dose in 51/324 of consecutively mobilised patients. Twenty three of these had >1x10Vkg GM- CFC, 22 proceeding to HOT with concomitant "back-up" bone marrow in two of these patients with very poor CD34+ yields of <0.5 x10®/kg. Neutrophil recovery was within 21 days but platelet independence was delayed (>28 days) in 8. Of the 28 patients with <1x107kg GM-CFC, 6 received HOT with PBSC alone and 5 had delayed engraftment. Twelve patients with <1x10Vkg GM-CFC received HOT supported by bone marrow collected after PBSC collection failure. Eleven patients were evaluable for engraftment and 4 had slow (>21 days) or delayed (>28 days) neutrophil recovery and 8 delayed platelet recovery. In the group of patients receiving <1x107kg GM-CFC there were 5 procedure related deaths. This study illustrates that delayed haematological recovery is frequent if <1x10Vkg CD34+ cells are infused after

HDT particulary with GM-CFC <1x10Vkg. The procedure related mortality in this latter group is high. In most patients whose PBSC harvest contains <1x10Vkg GM-CFC the use of bone marrow cells does not improve engraftment suggesting that poor PBSC mobilization usually indicates poor marrow function. It should be noted that poor PBSC yields are not always the result of impaired marrow stores. With some stem cell toxic mobilizing regimens for example, poor mobilization may not adversely affect marrow progenitor yields ((Dreger et al., 1995). In some instances poor yields may result from inappropriate mobilizing chemotherapy (Demirer & Bensinger, 1995), insufficient recovery from prior treatment (Perry et a/., 1998), or technical failures in harvest timing or apheresis procedures. If clinical circumstances permit it is probably preferable to allow at least 3 months recovery from chemotherapy and remobilize and harvest with an optimal protocol before assuming a mobilization failure.

In Chapter 7, one hundred and sixteen PBSC collections were subject to CD34+ cell purification using the CEPRATE® SC stem cell concentration system. The overall median purity of CD34+ cells was 70% (6-95%). CD34+ cell, and GM-CFC recoveries were 52% (8-107%) and 36% (3-118%). Purity was logarithmically related to the input percentage of CD34+ cells and starting requirements were established of 1% CD34 cell content for optimal purity and a minimum of 2 xIOVkg CD34+ cells to ensure recovery of our minimum engraftment threshold of 1 xIOVkg CD34+ cells. Reduction of the washing steps reduced non-specific cell losses and shortened the procedure but did not affect progenitor cell recovery. Purified CD34+ cells were reinfused following high dose therapy in 52 patients. The median time to neutrophil recovery of 0.5 x10®/L was 12 (9-23) days and to the attainment of platelet independence was 13 (7-100) days. The risks of delayed platelet recovery were related to the CD34+ cell dose infused and were identical to the risks when non-purified PBSC collections were used. Purification of CD34+ cells using the CEPRATE

device was shown to be reliable and the purified product resulted in prompt engraftment. The cell losses that occur do however restrict its use in many patients.

Haemopoietic recovery is more rapid after peripheral blood stem cell (PBSC) transplantation than after autologous bone marrow transplantation, and the aim of the study in Chapter 8 was to assess the role of the large number of lymphocytes and monocytes (accessory cells) in a PBSC leucapheresis product in this rapid regeneration. It has been noted by Mielcarek for example that PBSC harvests contain many more cells associated with cytokine production including at least 1-log more T-cells and 50-fold more monocytic cells (CD 14+) compared to a typical marrow graft (Mielcarek & Torok-Storb, 1997). Mielcarek provided in vitro evidence that sorted CD14+ cells from PBSC were associated with at least 10-fold higher levels of supernatent G-CSF and IL-6 cytokine production when incubated on long-term bone marrow stroma than equivalent marrow MNC and proposed that this may facilitate faster haematological engraftment with PBSC in vivo (Mielcarek et al., 1996). Higher plasma G-CSF and IL-6 levels post PBSCT compared to marrow autografts were noted by a number of other authors (Baiocchi at al., 1993; Ho at al., 1994; Kawano at al., 1993; Testa at al., 1994). Haematological recovery was therefore assessed in 10 PBSC recipients with lymphoma or myeloma in whom monocytes and T-cells were depleted by a median of 2.3 and 3.3 logs by CD34+ cell selection using the CEPRATE® SC stem cell concentration system and compared with recovery in 59 recipients who received whole PBSC. After allowing for the number of progenitor cells reinfused, there was no significant delay in engraftment induced by accessory cell depletion. Plasma levels of granulocyte-colony stimulating factor (G-CSF), granulocyte/monocyte-colony stimulating factor (GM-CSF), interleukin-6 (IL-6), stem cell factor (SCF) and macrophage-inhibitory pnitèin-alpha (MIP-1-alpha) during the transplant procedure were similar whether or not accessory cells were given. The G-CSF

and IL-6 levels rose between days 5 to 14 post transplantation to approximately 1 ng/ml and 50 pg/ml respectively. This study indicates that accessory cells reinfused with PBSC collections are not responsible for the subsequent cytokine profile or rapid haematological recovery. Intriguingly, in a recent in

vitro study by Gupta et al, purified GD34+ bone marrow cells were co-cultured

with irradiated stromal cells either in direct contact or separated by a transwell membrane and supernatant G-CSF and IL-6 concentrations measured. A significant 5-fold increase over baseline stromal supernatant levels to around lOOOpg/mL for IL-6 and 250pg/mL for G-CSF was observed in both culture conditions with CD34+ cells over a seven day period and suggests humoural stimulation of stromal cell cytokine production (Gupta et al., 1998). When more mature myelo-monocytic cells (CD15+CD14+) were incubated with the irradiated stromal cells supernatent IL-6 and G-CSF levels were also increased but only by 1.5-2-fold and transiently after 48hrs incubation. This study raises the interesting possibility that infused CD34+ cells in vivo may be directly responsible for a significant contribution to elevated cytokine levels.

The requirements of PBSC autografting are now well established and the work of these thesis contributes to the understanding of progenitor requirements, the feasibility and limitations of progenitor purification and addresses some of the issues of poor mobilization. This experience provides a platform which is essential to facilitate the major advances in graft engineering and cellular therapy anticipated in the next few years.

A notable feature of PBSCT has been its rapid clinical application based on the previous long experience with marrow transplantation. Despite some biological differences between these progenitor sources this has proved highly successful and durable for autografting (Haas et al., 1995). A more cautious approach is justifiable for the steadily rising replacement of marrow allografting with blood MPC however as these cells will provide the sole source of engraftment for the lifetime of the recipient.

Although there are at least 10-fold more T-cells infused in allogeneic PBSCT there has been no significant increase in acute GvHD reported and, as in the autologous setting, haematological recovery is significantly more rapid (Hagglund et al., 1998; Ringden eta!., 1998) with resultant savings of 30% in the cost of the first 100 days estimated in one study (Faucher et al., 1998). The greater number of stem cells possible with PBSC has also facilitated the use of so called low-intensity or "mini-transplants" in which chimerism and hopefully a graft versus tumour effect is established without fully myeloablative therapy (Slavin et al., 1998). This appears to be less toxic than conventional transplants and can therefore be used for patients previously excluded because of higher age or organ dysfunction for example (Giralt et al., 1997).

Despite these apparent advantages, the safety of the donor is of paramount concern and the adverse consequences of the mobilization and collection of PBSC must be weighed against previous experience with bone marrow donation. There have been two major studies addressing the risks of marrow