Study Design For Childhood Cancer Trials

PHASE I STUDY DESIGN

Because childhood cancer is rare and the response to conventional treatment good, most children never experience recurrent disease and are thus

Table 7.3. Risk group assignments for intergroup Rhabdomyosarcoma Study Group study V

Risk (protocol)

Stage

Croup

Site

Size

Age

Histology

Metastasis

Nodes

Treatment

Low, subgroup A

1

l

Favourable

a or b

<21

EMB

MO

NO

VA

(D9602)

1

ll

Favourable

a or b

<21

EMB

MO

NO

VA + XRT

1

III

Orbit only

a or b

<21

EMB

MO

NO

VA + XRT

2

l

Unfavourable

a

<21

EMB

MO

NO or NX

VA

Low, subgroup B

1

ll

Favourable

a or b

<21

EMB

MO

N1

VAC+XRT

(D9602)

1

III

Orbit only

a or b

<21

EMB

MO

N1

VAC+XRT

1

III

Favourable

a or b

<21

EMB

MO

NO or N1 or NX

VAC+XRT

(excluding

orbit)

2

ll

Unfavourable

a

<21

EMB

MO

NO or NX

VAC+XRT

3

I or II

Unfavourable

a

<21

EMB

MO

N1

VAC (+XRT, Cp II)

3

I or II

Unfavourable

b

<21

EMB

MO

NO or N1 or NX

VAC (+XRT, Cp II)

Intermediate

2

III

Unfavourable

a

<21

EMB

MO

NO or NX

VAC ± Topo + XRT

(D9803)

3

III

Unfavourable

a

<21

EMB

MO

N1

VAC ± Topo + XRT

3

III

Unfavourable

b

<21

EMB

MO

NO or N1 or NX

VAC ± Topo + XRT

1 or 2 or 3

I or II or III

Favourable or

a or b

<21

ALV/UDS

MO

NO or N1 or NX

VAC ± Topo + XRT

unfavourable

4

I or II or III

Favourable or

a or b

<10

EMB

M1

NO or N1 or NX

VAC ± Topo + XRT

or IV

unfavourable

High (D9802)

4

IV

Favourable or

a or b

>10

EMB

M1

NO or N1 or NX

CPT-11, VAC + XRT

unfavourable

4

IV

Favourable or

a or b

<21

ALV/UDS

M1

NO or N1 or NX

CPT-11, VAC + XRT

Favourable = orbit/eyelid, head and neck (excluding parameningeal), genitourinary (not bladder or prostate) and biliary tract. Unfavourable = bladder, prostate, extremity, parameningeal, trunk, retroperitoneal, pelvis, other, a = tumour size <5 cm in diameter; b = tumour size >5 cm in diameter.

EMB = embryonal, botryoid, or spindle-cell rhabdomyosarcomas or ectomesenchymomas with embryonal RMS.

ALV = alveolar rhabdomyosarcomas or ectomesenchymomas with alveolar RMS, LJDS = undifferentiated sarcomas.

NO = regional nodes clinically not involved; N1 = regional nodes clinically involved; NX = node status unknown.

VAC = vincristine, actinomycin D, cyclophosphamide; XRT = radiotherapy; Topo = topotecan; Gp = Group; CPT-11 = irinotecan.

Ed 73

Favourable = orbit/eyelid, head and neck (excluding parameningeal), genitourinary (not bladder or prostate) and biliary tract. Unfavourable = bladder, prostate, extremity, parameningeal, trunk, retroperitoneal, pelvis, other, a = tumour size <5 cm in diameter; b = tumour size >5 cm in diameter.

EMB = embryonal, botryoid, or spindle-cell rhabdomyosarcomas or ectomesenchymomas with embryonal RMS.

ALV = alveolar rhabdomyosarcomas or ectomesenchymomas with alveolar RMS, LJDS = undifferentiated sarcomas.

NO = regional nodes clinically not involved; N1 = regional nodes clinically involved; NX = node status unknown.

VAC = vincristine, actinomycin D, cyclophosphamide; XRT = radiotherapy; Topo = topotecan; Gp = Group; CPT-11 = irinotecan.

Source: Reproduced from Raney ef a/.31 (p. 21 8), with permission.

Table 7.4. International Neuroblastoma Staging System (INSS)

Stage 1: Localized tumour continued to the area of origin; complete gross resection, with or without microscopic residual disease; identifiable ipsilateral and contralateral lymph node negative for tumour.

Stage 2A: Unilateral with incomplete gross resection; identifiable ipsilateral and contralateral lymph node negative for tumour.

Stage 2B: Unilateral with complete or incomplete gross resection; with ipsilateral lymph node positive for tumour; identifiable contralateral lymph node negative for tumour.

Stage 3: Tumour infiltrating across midline with or without regional lymph node involvement; or unilateral tumour with contralateral lymph node involvement; or midline tumour with bilateral lymph node involvement.

Stage 4: Dissemination of tumour to distant lymph nodes, bone marrow, liver or other organs except as defined in stage 4S.

Stage 4S: Localized primary tumour as defined in stage 1 or 2, with dissemination limited to liver, skin or bone marrow

Risk group and protocol assignment schema: POG and CCG

INSS stage

Age (y)

N- myc status

Shimada histology

DNA ploidy

Risk group/study

1

0-21

Any

Any

Any

Low

2A and 2B

<1

Any

Any

Any

Low

>1-21

Nonamplifieda

Any

NA

Low

>1-21

Amplifiedb

Favourable

NA

Low

>1-21

Amplified

Unfavourable

NA

High

3

<1

Nonamplified

Any

Any

Intermediate

<1

Amplified

Any

Any

High

>1-21

Nonamplified

Favourable

NA

Intermediate

>1-21

Nonamplified

Unfavourable

NA

High

>1-21

Amplified

Any

NA

High

4

<1

Nonamplified

Any

Any

Intermediate

<1

Amplified

Any

Any

High

>1-21

Any

Any

NA

High

4S

<1

Nonamplified

Favourable

>1

Low

<1

Nonamplified

Any

1

Intermediate

<1

Nonamplified

Unfavourable

Any

Intermediate

<1

Amplified

Any

Any

High

aN-myc copy number <10. bN-myc copy number >10.

aN-myc copy number <10. bN-myc copy number >10.

POG, Pediatric Oncology Group; CCG, Children's Cancer Group; INSS, International Neuroblastoma Staging System; NA, not applicable. Source: Reproduced from Castleberry,60 (pp. 926, 930), with permission from Elsevier.

not eligible for trials of new agents. Phase I trials are designed to estimate the maximal tolerated dose of a drug, to determine the nature and frequency of toxicities, and to define the drug pharmacokinetics. While eligibility varies, patients have typically failed front-line therapy and usually they will also have failed second-line therapy. Because of the small number of pae-diatric patients eligible for Phase I trials, most are accomplished as multi-institutional collaborations. Paediatric drug development requires separate Phase I studies (i.e., separate and distinct from studies done in adults) because pae-diatric patients may tolerate either higher or lower levels of drugs and may exhibit toxici-ties unique to children. Separate trials warranting emphasis may also reflect unique agents active in paediatric tumours, differing from agents that are of the highest priority for cancers common among adults.

The basic design is to begin at about 80% of the adult maximal tolerated dose. Patients are entered in cohorts and treated at increasing doses. At each level, three patients are typically accrued. If there is no dose-limiting toxicity amongst the three patients, the dose is raised to the next level (usually a 20-30% escalation), in successive cohorts of patients with no intrapatient dose escalation. If two or all three of these initially accrued patients experience dose-limiting toxic-ity (DLT), the maximum tolerated dose (MTD) will have been deemed exceeded. Finally, if one patient amongst the initial three patients experiences dose-limiting toxicity, an additional three patients are accrued. If six patients are needed, a dose escalation will occur if a total of one in six (i.e. zero of the next three) has dose-limiting toxicity. If two or more (i.e. one or more of the next three) experience dose-limiting toxic-ity, the maximal tolerated dose will be deemed to have been exceeded. The MTD is defined as the dose level immediately below the level at which two patients in three to six experience DLT. The definition of dose-limiting toxicity can vary from study to study, but it generally falls into two categories: (a) Grade 3, 4 or 5 non-haematologic toxicity other than (1) Grade 3 nausea/vomiting; (2) Grade 3 transaminase elevation; and (3) Grade 3 fever/infection and (b) Grade 4 myelosupression, that lasts more than 7 days, which requires transfusions twice in 7 days, or causes a delay in therapy exceeding 14 days. While the study is temporarily closed after accrual of each set of three patients in order to assess patient-specific responses and toxicities, a patient reservation system is used to obtain places when and if the study reopens. Phase I trials often require the evaluation of many dose levels. At a given dose level, the probabilities of declaring that the MTD has been exceeded are 9.3%, (50%) and [83%], when the true probabilities of dose-limiting toxicities are respectively 0.1, (0.3) and [0.5].

Consensus guidelines established by American and European investigators for the conduct of paediatric Phase I trials have been established.32 A problem recently identified is the determination of MTDs in paediatric trials that are lower than those defined in adult patients, which may relate to differences in the intensity of prior therapy between adult and paediatric patients entered onto Phase I trials. There is a well-established association between prior therapy and reduced tolerance to myelotoxic drugs. If current paediatric Phase I trials in heavily pretreated patients define MTDs that tend to be lower than those determined in adult patients with minimal prior therapy, then application of the paediatric MTD to less heavily pretreated paediatric patients, e.g., in Phase II trials, may be problematic.

PHASE II STUDY DESIGN

The specific purpose of a Phase II trial is to determine activity, i.e., to develop estimates of the response rate of patients with specific tumour types to a particular drug or novel combination. Eligible patients typically will have relapsed on a front-line therapy, and the prospect of a cure is unlikely. Typically, the dependent variable is an objective all or none response variable such as achievement of a complete or partial (>50%) response. Interim results are masked from the participants until the study closes to accrual and response information for all patients has been established. There are three types of Phase II trial designs that depend upon the study objectives.

Testing Activity

The most common is 'proving activity'. For these studies, a fixed objective response rate is specified for activity (null hypothesis), and the goal is to reject the hypothesis in favour of the alternate hypothesis that the response rate is greater than this fixed figure. Generally, since the number of Phase II agents that can be tested is large in comparison to patient availability, sequential designs are preferred. However, as Simon33 pointed out, it is rarely advantageous to go beyond two stages. Two excellent references with regard to Phase II design are Simon33 and Shuster34 The designs of Simon33 stop at the first stage only if lack of activity is demonstrated. His argument is that patients should benefit from active drugs. However, in paediatrics, due to the relative scarcity of patients with recurrent disease, designs that stop early for either lack of activity or proven activity are preferred.

Historical Comparison

Another strategy for defining efficacy would be to prove a response rate is superior to that seen in an historical control study. The response rate of the new study is statistically compared to that of the control therapy. Makuch and Simon35 have provided methods to determine the sample size requirements for these studies. Chang et al?6 have extended this to two-stage designs (i.e., a sequential approach that could save patient resources).

Randomised Phase II Comparison

Due to a limited availability of patients, it is exceedingly rare that a randomised comparison of a new agent to a control is feasible in a paediatric Phase II study. However, such studies have been done. See McWilliams et al.37 for an example from childhood neuroblastoma. As above, two-stage or group sequential designs are the preferred method. The programme EAST38 can be used for designs that allow for both early acceptance and early rejection of the null hypothesis that the new treatment is equivalent to the control treatment.

In paediatric oncology, with limited patient numbers, only one or two cooperative Phase II trials are conducted with each new agent, and all malignancies refractory to standard therapy are typically combined into a single paediatric Phase II trial, usually stratified by histology. Not surprisingly, Phase II trials of novel multiagent regimens provide greater evidence of activity than single agent Phase II trials and offer considerable possibility of therapeutic benefit.39

PHASE III DESIGN

These studies typically ask a randomised question about either survival or event-free survival (the time from study entry to the earliest of induction failure, relapse, second cancer, or death of any cause). Intent-to-treat40 is the analysis of choice for efficacy, with other analysis done as secondary supportive inference. For treatment questions where the randomised divergence is considerably after study entry or where a significant number of failures are expected to occur before divergence, a delayed randomisation is typically done as close to the divergence point as possible. For these randomisations, the dependent variable would be event-free survival from the randomisation date.

Phase III studies are typically designed assuming either proportional hazards or the cure model of Sposto and Sather.41 In either case, the designs are group sequential in nature with planned interim analyses. In the case of proportional hazards, the O'Brien-Fleming method42 is used. The reader is referred to Shuster43 for specific details. Nearly all Phase III childhood cancer trials are run either as two-armed studies or as 2 x 2 factorial studies. It is rare that sufficient numbers of paediatric cancer patients are available to conduct three-armed studies, except perhaps in ALL, the most commonly occurring malignancy. The type of questions utilised in 2 x 2 factorial studies must be such that the expectation is for no 'qualitative interaction' between the two interventions. A qualitative interaction between treatments A and B would occur if a standard regimen plus A is superior to the standard regimen alone, but the standard plus A plus B is inferior to the standard plus B. For example, if a study is to randomise leukaemia patients to receive or not receive regimen A, designed to have an impact on the CNS, while at the same time to receive or not receive regimen B, designed to have an impact on marrow remission, a factorial design would seem appropriate. Essentially, we can run two studies for the price of one. If the two interventions have much in common, this would be a contraindication for a factorial design. In contrast, if we wished to ask if the same drug had an impact in induction therapy (first intervention) and in maintenance therapy (second intervention), there is, at least intuitively, the plausibility that the advantage of both interventions over just one may be zero or even harmful.

Phase III studies done in cooperative groups are required by the NCI to have a Data Safety and Monitoring Board which reviews the study at a minimum of every six months for toxicity and at planned intervals for efficacy, until it releases the study to the study committee. The release can occur no sooner than the earlier of (1) all subjects have completed the planned intervention or (2) the study was closed early and a new intervention is needed for patients on one or both arms. Any release prior to the planned date of final analysis requires approval of the board. Double-blind Phase III studies are rarely feasible due to the toxic nature of cancer treatment. However, they are encouraged for studies of supportive care, as long as the intervention is given in a pill form, and has no major known side effects requiring special medical monitoring.

Negative questions are often posed for paedi-atric cancer. For such studies, a very high cure rate of at least 85% has been shown possible on a conventional regimen. The question posed is can we do 'almost as well' with reduced therapy? To answer such questions with confidence requires large numbers, and it is rare that even the entire patient resources of COG are sufficient to address this in a randomised manner. For example, if a disease has a historical 4-year remission rate of 90%, and an accrual rate of 200 patients per year, a randomised study would take 6 years of accrual (10-year duration) to have 95% power to detect a degradation to 85% under reduced therapy at p = 0.20, one-sided. (Note that the typical values of type I error and power are reversed.) A singlearm study would require 315 patients to ask the same question of a fixed standard of 90% vs. a reduction to 85% (nearly a 75% reduction in sample size). While the benefits of reduced therapy may be obvious, such studies carry considerable risk and must be carefully monitored for early evidence that the reduction in therapy is unsafe and is associated with an inferior outcome.

ANCILLARY STUDIES

In paediatric cancer, there is considerable activity in translational research (see above). This can take the form of biologic studies, late effects, or in controlling acute side effects. These studies are designed on a case-by-case basis. Examples include the conduct of case-control 'tissue bank' studies to establish a promising prognostic marker. Cases are defined as patients failing a protocol (typically a relapse) and controls are long-term successes. These studies can be done using sequential designs, typically two-stage designs. Other typical studies might look at cognitive impairment (multivariate analysis of variance of neuropsychological variables), acute toxicity of a specified type (typical Chi-square test), the prognostic significance of serial pharmacologically measured drug levels (time-dependent covariate in survival analysis), or exploratory analysis (e.g. microarrays).

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment