Malignant Solid Tumors (Key Questions 1 and 2)

Evidence suggesting benefit of HSCT compared with conventional therapy:

• Low strength evidence on overall survival suggests a benefit with single HSCT compared to conventional therapy for high-risk recurrent or progressive anaplastic astrocytoma.

Evidence suggesting harm of HSCT compared with conventional therapy:

• Low strength evidence on overall survival suggests harm due to higher treatment-related mortality with single HSCT compared to conventional chemotherapy for nonanaplastic mixed or unspecified ependymoma.

Evidence suggesting no benefit of HSCT compared with conventional therapy:

• Moderate strength evidence on overall survival suggests no benefit with single HSCT compared to conventional therapy for metastatic rhabdomyosarcoma.
• Low strength evidence on overall survival suggests no benefit with single HSCT compared to conventional therapy for extraocular retinoblastoma with CNS involvement, high-risk Ewing's sarcoma family of tumors, and high-risk relapsed Wilm's tumor.

Insufficient evidence:

• The body of evidence on overall survival with tandem HSCT compared to single HSCT is insufficient to draw conclusions for high-risk Ewing's sarcoma family of tumors, neuroblastoma, central nervous system embryonal tumors, and pediatric germ cell tumors.
• The body of evidence on overall survival with single HSCT compared to conventional therapy is insufficient to draw conclusions for central nervous system embryonal tumors, high-risk rhabdomyosarcoma of mixed stages, congenital alveolar rhabdomyosarcoma, cranial parameningeal rhabdomyosarcoma with metastasis, allogeneic transplantation for metastatic rhabdomyosarcoma, extraocular retinoblastoma with no CNS involvement, trilateral retinoblastoma, and six types of glial tumor (newly diagnosed anaplastic astrocytoma, newly diagnosed glioblastoma multiforme, anaplastic ependymoma, choroid plexus carcinoma, recurrent/progressive glioblastoma multiforme, and nonanaplastic, mixed or unspecified ependymoma).

Nonmalignant Diseases: Inherited Metabolic Diseases (Key Questions 3 and 4)

The inherited metabolic diseases were split into three categories for this review. Rapidly progressive disease was defined as progression to death within 10 years; the outcome of interest is overall survival. Slowly progressive disease was defined as progression to death of 10 years or greater; the outcomes of interest are neurocognitive and neurodevelopmental outcomes. For diseases that have both rapidly and slowly progressive forms of disease, outcomes of interest are overall survival and neurocognitive and neurodevelopmental outcomes respectively.

Autoimmune Diseases (Key Questions 5 and 6)

The main consideration in this systematic review was the comparative balance of long-term benefits and harms of HSCT. With the exception of newly diagnosed type I juvenile diabetes, children in the studies reviewed herein had severe, typically disabling disease, refractory to a wide variety of standard therapies. Thus, the disease natural history in those settings assumed the role of comparator.

Insufficient evidence:

• The overall body of evidence is insufficient to draw conclusions about the comparative benefits (e.g., increased overall survival) or harms (treatment-related mortality, secondary malignancies) of single autologous or allogeneic HSCT versus conventional therapy or disease natural history in patients with newly diagnosed type 1 diabetes mellitus, or those with severe, refractory, poor prognosis autoimmune diseases, including: systemic lupus erythematosus, juvenile idiopathic arthritis, systemic sclerosis, malignant multiple sclerosis, Crohn's disease, myasthenia gravis, overlap syndrome, diffuse cutaneous cutis, Evans syndrome, autoimmune hemolytic anemia, and autoimmune cytopenia.
• Although the overall body of evidence is insufficient to come to conclusions about the relative balance of benefits (e.g., increased overall survival) or harms (treatment-related mortality, secondary malignancies), moderate strength evidence suggests that extended periods of drug-free clinical remission can be achieved in some cases with single autologous HSCT for patients with newly diagnosed type I juvenile diabetes, and severe, refractory juvenile idiopathic arthritis, systemic lupus erythematosus, systemic sclerosis, and Crohn's disease.

This systematic review addresses a broad range of diseases, for the majority of which HSCT is considered only in patients who have diseases with very poor prognosis, refractory to best available treatment. It is only in such settings that the rigors and risks associated with HSCT would likely be considered. These risks include treatment related mortality, iatrogenic infections secondary to neutropenia, potential for secondary malignancy and over the long term cognitive and developmental delays. Families and their physicians face not only the challenges of severe disease, but when these diseases are uncommon or rare, challenges to the accumulation of knowledge about effective therapy are substantial. The present systematic review offered the opportunity to rigorously assess the evidence for HSCT in pediatric disease; simultaneously, it revealed gaps in the evidence, suggesting opportunities to address these.

Cancer research has numerous well-defined conventions for reporting outcomes, but these were not used consistently in the literature. For example, overall survival may be reported as time from diagnosis for newly diagnosed disease or time from recurrence for relapsed disease. When reporting overall survival, papers were often unclear which time point was used in their calculation; some even reported overall survival from time of transplant. Moreover, some papers did not report overall survival at all, but reported only measures related to disease progression. Similarly there was lack of consistency in reporting adverse events. For example, even such an important harm as treatment related mortality was not always reported. Without an explicit statement of the occurrence of treatment-related mortality, it is impossible to ascertain if there was no mortality or a failure to report the mortality that occurred. Inconsistencies also occurred in the reporting of toxicities, although there are well-defined conventions for grading the severity of toxicity.

There were few randomized controlled trials for any of the indications included in the systematic review. While this might be expected with uncommon and rare diseases, some solid tumors reviewed herein had substantial numbers of patients. For example, some 600 patients underwent tandem HSCT for neuroblastoma. In high-risk Ewing's sarcoma and high-risk rhabdomyosarcoma, more than 250 patients underwent HSCT for each disease. The widespread reporting of aggregated data is another obstacle to evaluating the outcomes of HSCT. For example, it is common to report studies of HSCT that include patients with a variety of diseases without reporting disease specific outcomes. Within a single study, patients with disparate prognosis may be aggregated without reporting stratified results.

The inherited metabolic diseases illustrate how rare the disease, and thus the evidence, can be. Among those included in the systematic review, evidence typically consisted of no more than six cases. Yet some diseases have a homogeneous and dismal natural history. In particular, among diseases we classified as rapidly progressing, or refractory to standard therapies, spontaneous remission is highly unlikely or impossible. Therefore, we attributed the reported results to HSCT. For example, the implications of transplantation for a rapidly progressing lysosomal storage disorder like Wolman's syndrome are clear; this is a choice between certain death and potential survival, albeit with associated risk of adverse effects associated with transplant.

Most autoimmune diseases in children are rare, and particularly in the cases included in this report, represent a daunting therapeutic challenge. With the exception of newly diagnosed type I juvenile diabetes, patients with autoimmune diseases reviewed here had severe, disabling illness that had not responded to or had relapsed following a large number of standard therapies. HSCT was essentially a last resort for these children and adolescents. In a large proportion of subjects in those settings, HSCT was followed by a period of sustained remission of severe symptoms and therefore respite from immune suppressive therapy. While the durability of clinical remission, and the balance of long-term risks and benefits remains unknown, the obvious strength of association permits the conclusion that HSCT was likely causative.

By contrast, type I autoimmune juvenile diabetes can be satisfactorily managed over the long term, at relatively low risk, in a large proportion of children with intensive insulin therapy (IIT) and lifestyle modifications. The risk-benefit ratio for HSCT compared to IIT must balance the potential for long-term benefit (cure) and harms, particularly those associated with cytotoxic chemotherapy agents used in preparation for HSCT. While evidence suggests a sustained period of insulin independence and adequate metabolic control may be achieved with HSCT, the decision to apply this high-risk procedure to this population is not clear-cut. To date, no trials of HSCT in newly diagnosed type 1 diabetes have been conducted or registered in the U.S.

Future Research

The available literature to assess the comparative effectiveness of HSCT to conventional therapy in pediatric patients largely comprised small case series and case reports. The challenges of conducting research in rare diseases or rare disease settings need to be acknowledged. Many of these diseases are very rare, so the pace of patient accrual may be slow; this may be accompanied by changes in practices, both for induction chemotherapy and stem cell transplantation itself and other aspects of management and treatment. Also, patients are likely to be clinically diverse in terms of disease site, tumor histology or stage, prior and co-interventions, and other factors. Specific recommendations for future research follow.

1. For diseases with adequate patient populations, promote multicenter randomized trials to increase the scientific rigor in which HSCT can be evaluated.
2. Use established registries to standardize the collection of demographic data, treatments, and to facilitate the evaluation of comparative harms and benefits of treatments.
3. Recognizing that observational studies, including case series, and case reports will continue to be attractive to investigators, recommendations to improve the usefulness and generalizability of such studies are:
   • Conduct prospective studies with contemporaneous treatments.
   • Patients in both single arm and comparative studies would be comparable in terms of key variables, such as disease, anatomic site, disease stage, and prior treatment.
   • Consistent reporting of survival outcomes, with a clear definition of the survival time i.e., time from diagnosis, time from transplant or time from recurrence.
   • Consistent harms reporting is essential in facilitating the comparative evaluation two treatments. Complete reporting of treatment related mortality, secondary malignancy, serious infections, and veno-occlusive disease would be standard.
   • Make studies comparative when possible.
   • Multivariable regression analyses can be helpful in controlling for potential confounders, when sufficient sample sizes can be obtained, and would adhere to good modeling practices.^686-692
   • Guidance for study quality in observational studies has been addressed by Deeks et al.⁶⁹³
4. For solid tumors, future studies would focus on single diseases, and collect detailed information on prognostic factors that may allow for more refined stratification of high-risk categories which may highlight those likely to benefit from HSCT and allowing for less uncertainty in the interpretation of results. Followup would be sufficient to assess the impact of HSCT on the development of secondary malignancies and long term impact on neurocognitive development and fertility.
5. For pediatric patients with slowly progressive forms of inherited metabolic diseases, controlled trials with sufficient followup are needed to evaluate the long-term balance of benefit and harms. Trials would use standardized measure of neurocognitive and neurodevelopmental outcomes.
6. For pediatric patients with autoimmune diseases, controlled trials with sufficient followup are needed to evaluate the long-term balance of benefit and harms.