The safety and feasibility of stereotaxic implantation of two or six million hNT cell doses (60 and 180 µL, respectively) was initially demonstrated by Kondziolka and colleagues [14] and confirmed in the subsequent randomised trial, involving the stereotaxic implantation of five or ten million hNT cells in 250 µL total volume, in 25 patients with subcortical stroke [15]. Intracerebral delivery has also been described in stroke for fetal porcine cells, and autologous bone marrow-derived cells [36, 37]. Trials of primary fetal cell implantation for Huntington’s or Parkinson’s diseases include cell injections targeted to the basal ganglia of 0.5 up to 10 million cells in volumes of 10–200 µL.

Stereotaxic intraparenchymal delivery offers the advantage that a large and controlled number of cells are delivered adjacent to the site of ischaemic tissue damage, and circumvents the blood–brain barrier. Delivery by intraparenchymal injection imposes anatomical restrictions since some sites cannot be safely implanted (e.g. the brainstem), and carries a well-characterised risk of intracranial bleeding and seizures in the range of 1–2 % [38] and 2.4 % [39] respectively. In addition, there are procedural risks associated with general anaesthesia, hospitalisation and neurosurgery, all in a predominantly elderly population likely to be taking antithrombotic or anticoagulant treatment for stroke prevention.

Allometric scaling suggests that a dose equivalent to the efficacious dose in rats is approximately 20 million cells in humans. While non-clinical studies have not indicated the dose ceiling above which there is no further increase in efficacy, the volume of material that can be injected into the brain suggested that the highest practical dose of CTX in man for stroke is 20 million cells.

A conservative approach for the starting dose and dose escalation was used, as appropriate for a first-in-man trial. Previously, cell dosing in the brain for treatment of neurological disease has employed small doses at multiple sites in the brain. A major consideration that is specific to intracerebral delivery is the volume of injection since this alone may lead to adverse events through compression of brain parenchyma. The maximum implantation volumes delivered into the brains across three animal species (NOD-SCID mouse, rat and cynomolgus monkey) in the non-clinical development of CTX were compared to the proposed first-in-man volume in order to evaluate the safety margin on volume of delivery in humans. In the rat studies the volume of the implantation was ~ 1/111th of the rat brain volume. The volume of the proposed first-in-man dose is 40 µL, equivalent approximately to 1/32500th of the volume of the human brain. This represents a large safety margin of 300x in the proposed clinical trial compared to the rat studies (83x and 667x compared to cynomolgous monkeys and NOD SCID mice, respectively). In the two clinical trials of the hNT cell line, doses of 2–10 million cells were used, delivered by multiple needle trajectories and multiple injection boluses of 10–20 µL each at points along each needle pass. The initial dose volume in PISCES (40 µL in three patients) was lower than the starting dose volume of 60 µL in four patients in the first-in-man hNT trial.

Since the concentration of CTX drug product is fixed at 50,000 cells/µL, a dose of two million cells requires 40 µL and a dose of five million cells 100 µL total volume. A single needle pass was used for the 2 and 5 million cell doses, with a second needle pass in close proximity to the first for the 10 million cell dose (total volume 200 µL) and four needle passes (total volume 400 µL) for the 20 million cell dose. Doses were allocated serially so that successive groups received 2, 5, 10 and 20 million cells. Only one patient was treated at a time.

Screening was conducted over approximately 8 weeks before surgery to ensure that eligible patients had stable functional deficits as measured by the National Institutes of Health Stroke Scale (NIHSS), a measurement of neurological deficit that is widely used to characterise the severity of stroke in the acute phase. NIHSS score predicts survival and functional recovery [40–42], and deterioration by 4 or more points on the total NIHSS score is a widely used threshold to identify clinically significant deterioration. NIHSS was therefore used to monitor clinical status before and after implantation.

In addition to standard biochemistry and haematology panels, participants were screened for the presence of specific antibodies to CTX HLA antigens before and at multiple time points after cell implantation.

Structural magnetic resonance imaging (MRI) scans were obtained pre-implantation and at 3, 12 and 24 months to specifically seek evidence of haemorrhage, new infarction, inflammation or tumour growth at the site of cell implantation.

All implanted patients were identified to a national registry that provides lifelong surveillance for events such as cancer and death, so long as they remain resident in the same country. Prospective consent for postmortem examination of brain tissue was also sought at trial entry.

An independent data and safety monitoring board (DSMB) reviewed all clinical, imaging and safety data throughout the study. A decision to continue dosing at each dose level followed satisfactory review of the 28-day safety data for the first patient at that dose level; and to increase the dose to the next level following satisfactory review of the 3-month safety data for the last patient in the lower dose group.

As a safety trial, a control group was not included. Inclusion of a small number of non-operated control patients would be highly unlikely to be informative given the heterogeneity of stroke patients, would significantly slow recruitment, and randomisation to non-implantation may be unacceptable to patients [43]. Logically, the appropriate control group would undergo all procedures (including general anaesthesia and craniotomy) except cell implantation, but the invasive nature of this treatment in a population with potential morbidity from such procedures, was not felt to be justified in a phase 1 study.

As a first-in-human trial and with no reproductive toxicology studies conducted with CTX cells to date, only male stroke patients were included. In addition, there is theoretical concern that if tamoxifen is required at a later stage, it might “switch” the CTX cells on again causing them to proliferate.

Implantation of CTX cells within nerve cell clusters near but not directly within the lesion was planned, consistent with intrastriatal implantation in preclinical models. Ischaemic lesions from stroke occurring within the internal capsule can interrupt large numbers and types of nerve fibres, including cerebral cortical efferents to the basal ganglia, thalamus, brainstem and spinal cord. In addition, afferent fibres projecting from the thalamus to the cortex can be interrupted. There is considerable variability between patients in the number and type of fibre affected, depending upon the size and specific location of the stroke within the white matter. It is considered hazardous to inject directly into the white matter lesion, as pressure resulting from the injection could cause further injury of the already damaged tissue. In seeking a neuronal cluster near most white matter ischaemic lesions, the putamen was selected as the best target, being large and easily accessible with stereotaxic approaches, and in close proximity to the site of many middle cerebral territory strokes.

The timing for cell implantation in PISCES was selected to ensure the recruitment of patients with stable neurological and functional deficits. Studies have found consistently that functional recovery from disabling stroke reaches a maximal level by 6 months [44–48], although the great majority of such studies must be qualified by recognising that both speed and completeness of apparent recovery are dictated by the chosen measure, which may be insensitive to deficits that are nonetheless significantly disabling; the BI has well described ceiling effects, for example [49]. In addition, the setting of previously reported studies may introduce bias, as the majority were conducted in specialist rehabilitation services, referral or eligibility for which may be dictated in part by perceived rehabilitation potential. In the Copenhagen Stroke Study [44, 45], recovery of neurological function was complete within 12.5 weeks from stroke onset in 95 % of patients and time course of recovery was clearly related to initial stroke severity. The best recovery of ADL occurred within 8.5 weeks for mild stroke, within 13 weeks for moderate stroke, within 17 weeks for severe stroke and within 20 weeks for very severe stroke, with no further significant improvement after this point. In another study, half of the patients with disabling ischaemic stroke recovered within 18 months although recovery was greatest in the first 6 months [50]. A minimum delay of 6 months therefore ensures that spontaneous recovery of function is highly unlikely, and that participants are likely to be medically stable.

Transplantation of tissue or cell-based products is usually combined with immunosuppression to guard against rejection. In respect of CTX, non-clinical studies have shown that cell survival and associated beneficial effects are not influenced by treatment of the animals with immunosuppressive drugs. Furthermore, in vitro studies for MHC-DR and MHC-ABC showed that expression of MHC class l and class II protein levels was low in CTX cells, indicating that the potential for rejection was low. Infections commonly complicate the poststroke period, especially pneumonia and urinary tract infections, and are strongly associated with poor outcome [51]. A prolonged period of immunodepression is recognised after stroke [52] and is likely to increase vulnerability. Immunosuppressive therapy was therefore considered to have significant risks, without evidence of either need or avoidance of complications.

Observations from the first PISCES trial support further studies, which will begin to explore efficacy. Initial experience permits some relaxation of recruitment criteria relating to initially cautious safety assessment, such that the lower age limit for the next stage of trials has been lowered to 40 years, and women are also included. Immunosuppression is again not planned. Earlier administration of cells than in PISCES will be undertaken since the mechanistically relevant recovery processes are most prevalent in the initial few weeks after stroke, which is also most consistent with the data from preclinical studies. For intraparenchymal administration at late time points after stroke, the blood–brain barrier might prevent systemic immune reactions and rejection, but this may not be so if subacute administration is undertaken, when the blood–brain barrier is compromised in the early stages of the ischaemic injury.