Introduction

In the USA, cancer is the second leading cause of death in children. Over the past 30 to 40 years, the survival of children with cancer has improved markedly (Table 1). The 5-year overall survival (OS) rate was 58 % from 1975 to 1977; this contrasts to the 5-year OS rate of 83 % from 2001 to 2007 [1]. Because more children are surviving from cancer, attention to the late effects of treatment has become more important. Recent studies have shown that about 25 % of deaths in 5-year survivors of childhood cancer can be attributed to treatment [2].

Table 1 The 5-year overall survival (in percent) of patients (ages 0 to 14 years) with childhood cancer over time based on the Surveillance Epidemiology and End Results review [1]
Full size table

Surgery, chemotherapy and radiotherapy (RT) have all been associated with different late toxicities. RT has been shown to be a major contributor in the development of late toxicity from treatment [2, 3]. While RT can cause inflammatory and fibrotic reactions in both children and adults, some late complications are seen only in children because RT can impair the growth and development of normal organs [4]. Furthermore, younger patients are more susceptible to development of secondary cancers [3]. Over the past years, there has been emphasis on minimizing the use of RT in many pediatric tumors. Several strategies have been employed in pediatric oncology to minimize the late effects of RT. These have included (1) omitting or delaying RT until the child is older, (2) decreasing RT doses and volumes by incorporating chemotherapy in the treatment regimen, (3) alteration of RT fractionation, (4) use of novel techniques to spare or minimize RT dose to surrounding normal tissues, (5) elimination of RT in favorable subsets of patients, and (6) reduction of RT volume through analysis of patterns of failure. This review outlines the different strategies that have been used including the published results of these treatments.

Omitting or delaying RT in the very young

Infants are the most susceptible to RT late complications such as cognitive deficits and growth problems. During the first few years of life, the brain, muscle, and bones are in a phase of rapid growth which slows down at around 7 years. A second peak of growth occurs in the musculoskeletal system at puberty. Consequently, some oncologists have advocated omitting or delaying RT in the very young until the child is older.

A popular approach three to four decades ago was delaying RT in infants with malignant brain tumors. Children less than 3 years of age underwent a maximal safe resection followed by 12 to 24 months of chemotherapy until the child was 3 years of age or until progression occurs. The results of this approach have been inferior compared with studies with older patients who are able to undergo craniospinal or cranial RT. The Pediatric Oncology Group infant study showed a 1- and 2-year progression-free survival of 42 and 34 % in medulloblastoma and 58 and 42 % in ependymoma using two courses of vincristine and cyclophosphamide alternating with one course of etoposide and cisplatin [5]. The French Society of Pediatric Oncology (SFOP) has reported a 4-year progression-free survival of 22 % in ependymoma after resection and postoperative chemotherapy [6]. In an analysis of the Surveillance Epidemiology and End Results database, children <3 years of age with ependymoma who receive immediate postoperative RT have a better 3-year OS compared with those not receiving postoperative RT (81 vs. 56 %, p = 0.005) [7]. A report from St. Jude Children’s Research Hospital showed a 3-year progression-free survival of 74.7 % in 88 children with a median age of 2.9 years after postoperative RT for ependymoma. Neurocognitive testing revealed stable findings, with more than half of the children tested at or beyond 2 years [8]. A strategy of omitting RT in medulloblastoma patients was performed in Germany using three cycles of cyclophosphamide, vincristine, methotrexate, carboplatin, and etoposide in addition to intraventricular methotrexate. The 5-year progression-free and OS rates were 82 and 93 % for the 17 patients who had complete resection, 50 and 56 % for those with residual tumor in the postoperative bed and 33 % and 38 % for those with macroscopic metastasis. A surprising finding in the study is the better outcome of patients with desmoplastic compared with classic medulloblastoma [9].

Infants <1 year old with rhabdomyosarcoma have a worse survival outcome compared with older patients primarily because of a higher local failure rate secondary to omission of RT. A report from the Italian Cooperative Group showed a local failure rate of 54 and 28 % in those who did not have and those who had RT when there was an indication for treatment [10]. Investigators from Memorial Sloan-Kettering Cancer Center have shown a 2-year local control of 84 % with RT in children <3 years of age with rhabdomyosarcoma. Mild functional deficits were seen using primarily a conformal RT approach; however, follow-up is short to determine the true extent of late toxicity [11].

Use of chemotherapy to lower RT dose and volume

In certain brain tumors, chemotherapy has been used to decrease RT volume and dose. In children with standard-risk medulloblastoma, 23.4 Gy craniospinal RT followed by posterior fossa boost and adjuvant chemotherapy seem to have an equivalent and maybe better survival compared with those who have received 36 Gy craniospinal RT followed by posterior fossa boost without adjuvant chemotherapy [12, 13]. There is also evidence that the tumor bed with a margin can be safely treated to a higher dose instead of the entire posterior fossa when chemotherapy is used [14]. Neurocognitive studies have shown that patients receiving 23.4 Gy craniospinal RT had fewer declines in mean intelligence quotient (IQ) scores compared with those receiving 36 Gy craniospinal RT in both younger (<7 years) and older (≥7 years) patients [15]. Another intracranial tumor where chemotherapy may impact on RT fields and doses is germinoma. For decades, the standard treatment for intracranial germinoma has been craniospinal RT. Some studies have shown that whole brain and whole ventricular RT fields followed by a boost may suffice in patients with M0 or non-metastatic disease. More recently, neoadjuvant chemotherapy has been used to shrink the tumor and tailor RT fields and doses [16]. A Children’s Oncology Group (COG) study randomizing patients to RT alone versus neoadjuvant chemotherapy followed by RT recently closed because of poor accrual. In patients with a complete response (CR) after neoadjuvant chemotherapy, RT is given to the primary site only and given 30 Gy. Patients who do not obtain a CR or who are randomized to RT alone receive whole ventricular RT to 24 Gy followed by a boost to the primary site for a total dose of 45 Gy. The SFOP published their results using four cycles of chemotherapy followed by 40 Gy RT to the primary site alone, regardless of chemotherapy response; 10 of 60 patients relapsed. Eight of the ten relapses were periventricular, suggesting that whole ventricular region should have been treated [17].

In stage III favorable histology Wilms’ tumor, the addition of doxorubicin to actinomycin-D and vincristine has allowed a dose reduction from 20 to 10 Gy flank or abdominal RT [18]. A study from the University of Iowa has shown lower proportion of patients with scoliosis by using less RT dose [19]. In Hodgkin lymphoma, children have traditionally been treated with subtotal nodal RT alone to doses of 40 to 44 Gy with subsequent musculoskeletal growth abnormalities [20]. In the past three decades, this approach has been challenged in children, where the current management is to deliver multiagent chemotherapy followed by involved-field RT to doses of 15 to 25.5 Gy with similar survival outcomes [21, 22]. The reduction of RT fields and lowering of dose has reduced the amount of breast tissue receiving radiation. Likewise, retardation of bone growth which seems to be more prominent above 25 Gy should also be reduced with this approach. Studies from the Netherlands Cancer Institute and Stanford University have conflicting results regarding the reduction of secondary breast cancer following involved field RT [23, 24]. A study from St. Jude Children’s Research Hospital has shown that even with a dose of 15 Gy, growth of the clavicle can still be impaired after asymmetric mantle RT, more prominent in younger children [25].

Alteration of RT fractionation

Hyperfractionation delivers a lower dose per fraction given more than once a day. Because the late effects of RT are directly related to the fraction size, lowering the dose per fraction, in theory, should reduce late complications. Retrospective analysis of musculoskeletal complications after hyperfractionated RT for Ewing sarcoma at the University of Florida has shown less fracture, loss of range of motion, loss of muscle circumference and atrophy in patients receiving hyperfractionated compared with conventional RT [26]. A review of patients treated with hyperfractionated RT for medulloblastoma at the New York University has shown a lower incidence of hypothyroidism in those receiving hyperfractionated compared with conventional RT [27].

There are at least five randomized trials in pediatric radiation oncology comparing conventional fractionation to hyperfractionated RT (Table 2). Some have used a hyperfractionated regimen to increase dose and local tumor control while keeping the same late effects as conventional RT [28–31]. In one of the protocols, RT was a hyperfractionated split-course regimen [30]. Others have given the same dose in a hyperfractionated manner with the aim of reducing long-term toxicity [32]. In all of these trials, there has been no documented improvement as far as local control or late toxicity with hyperfractionated RT.

Table 2 Randomized trials in childhood cancer comparing conventional to altered fractionation radiation therapy
Full size table

Use of novel RT technology

In the last two decades, many advances have occurred in RT treatment delivery. Three-dimensional conformal radiotherapy (3D-CRT) has been shown to improve target volume coverage as well as sparing of surrounding, normal tissues from high dose-RT when compared with conventional RT [33, 34].

Intensity-modulated radiation therapy (IMRT) has also been used in the treatment of children with brain, head and neck and soft tissue tumors. The local control rates with IMRT have been shown to be equivalent if not better than conventional RT in children (Table 3) [14, 35–43]. A comparison of 3-DCRT and IMRT in pediatric nasopharyngeal carcinoma showed no difference in the 2-year locoregional control but less grade 3 acute toxicities of the skin, mucous membrane and pharynx with IMRT [37]. In COG D9803, the 5-year locoregional failure rates were 18 % for 3D-CRT and 15 % for IMRT. While the locoregional control rates were not statistically different, target dose coverage was better for IMRT [42]. Regarding late toxicity, the main complication that IMRT might have an advantage compared with more conventional techniques is ototoxicity in children receiving RT and cisplatin for medulloblastoma [35, 44, 45]. The initial paper from The Methodist Hospital of 15 patients with a median follow-up of 18 months demonstrated a reduction from 64 to 13 % grades 3 to 4 hearing loss with lowering of RT dose to the cochlea [44]. A more recent analysis with 44 children and a median audiogram follow-up of 41 months showed 25 % grades 3 to 4 hearing loss [45]. Another report from Memorial Sloan-Kettering Cancer Center showed a 6 % grade 3 hearing loss with IMRT at a median audiogram follow-up of 19 months [35]. While increased conformality of the high-dose region to the target is the main advantage of IMRT, there have been several potential disadvantages with the use of this technique [46]. There is increased dose to non-target tissues secondary to increased low-dose radiation to surrounding critical structures otherwise not delivered with conventional RT from complex beam arrangements as well as increased low-dose radiation to the rest of the body secondary to increased monitor units to deliver desired dose to the target and leakage radiation. This technique, like many of the newer techniques of treatment delivery, requires prolonged treatment planning and delivery time and in some cases, prolonged anesthesia time in the very young.

Table 3 Local control outcome in children with brain, head and neck, and soft tissue tumors treated with intensity-modulated radiation therapy
Full size table

More recently, the use of proton therapy has gained wide acceptance in the pediatric oncology community. Because of the physical characteristics of the proton beam, there is a theoretical advantage for its use because of the reduction of integral dose to normal tissues when compared with photons. The ability of protons to spare normal tissues beyond a specified depth results in lower or no dose to surrounding organs. Since there are only a few centers in North America, Europe, and Asia using this type of treatment, information regarding clinical outcome is limited and still accumulating. The available literature suggests that local control with proton therapy is comparable to results obtained with IMRT and 3D-CRT; however, the number of patients treated is small, and follow-up is short (Table 4) [47–56]. Whether decreased late toxicity can be achieved by using protons remains to be seen. A recent prospective study of health-related quality of life for children with brains tumors treated with protons showed that at 3-year follow-up, patients treated with protons had very close quality-of-life scores compared with healthy children and better than children with chronic illness [57]. A review of 19 children treated with proton therapy for medulloblastoma showed grades 3 and 4 ototoxicity in 5 % at 1 year follow-up [58]. In another study of brain tumor patients, 47 % receiving protons alone and 33 % of children receiving protons in combination with photons developed endocrine dysfunction [59].

Table 4 Local control outcome in children with brain and extracranial solid tumors treated with protons
Full size table

Elimination of RT in favorable subset of patients

In children with favorable outcomes, one of the strategies has been to determine whether RT can be eliminated in the treatment regimen. In children with stage II favorable histology Wilms’ tumor, the National Wilms Tumor Study-3 randomized patients to 20 Gy RT vs. no RT. In addition, there was a second randomization to vincristine and dactinomycin with or without doxorubicin. Patients who received vincristine and dactinomycin without RT after nephrectomy did very well, eliminating the need for abdominal or flank RT in this favorable subset of patients [18]. Currently in COG AREN0533, the value of whole lung RT in favorable histology Wilms’ tumor is being studied. In children with lung metastasis who achieve a pulmonary CR after 6 weeks of vincristine, dactinomycin, and doxorubicin, whole lung irradiation is omitted if the only site of metastasis is the lung and there is no loss of heterozygozity at 1p and 16q.

In Hodgkin lymphoma, the elimination of involved-field radiotherapy (IFRT) is associated with a slightly worse event-free survival (EFS) but no difference in OS because of effective salvage therapy. Trials have looked at tailoring treatment based on response to initial chemotherapy (response-adapted therapy). The Children’s Cancer Group 5942 study showed a 10-year EFS and OS of 91.2 and 97.1 % for those receiving IFRT and 82.9 and 95.9 % for those not receiving IFRT after an initial CR to chemotherapy. For EFS and OS comparisons, p = 0.004 and 0.5, respectively [60]. The German study GPOH-95 also showed a worse EFS in children not receiving IFRT in the intermediate (stage IIEA, IIB, and IIIA) and high-risk groups (stage IIEB, IIIEA, IIIB, and IV) without a difference in OS when compared with those who received IFRT [61]. Current COG protocols (AHOD0431 and AHOD03P1) are examining which patients can RT be safely omitted.

Another group of patients with a favorable outcome are those with intracranial germ cell tumors. An international cooperative trial delivered four cycles of carboplatin, etoposide, and bleomycin to patients with intracranial germ cell tumors [62]. Those who had a CR received an additional 2 cycles of the same agents while those who didn’t had intensification with cyclophosphamide. None of the patients had RT as initial part of therapy. A total of 78 % of patients had a CR without irradiation. With a median follow-up of 31 months, 39 % were alive without relapse or progression of disease. Salvage therapy was successful in 93 % of patients who relapsed or progressed.

In children with orbital rhabdomyosarcoma, the standard approach by the Intergroup Rhabdomyosarcoma Study is to deliver RT for group III disease. In Europe, RT is sometimes omitted because of late toxicity in these young patients. A previous report revealed that only 37 % of patients received RT as primary local treatment in SIOP. The 10-year EFS rates for patients receiving RT and no RT were 82 % and 53 % (p < 0.001). The 10-year OS rates were not statistically different (88 % and 85 % respectively) [63]. Orbital hypoplasia and cataracts were more common in those who had RT.

While in Hodgkin lymphoma, intracranial germ cell tumor and orbital rhabdomyosarcoma, OS is not different with omission of RT, local failure is higher as well as use of salvage therapy. These issues need to be taken into account in treatment decision making. On one hand, late toxicity is higher in children receiving RT. On the other hand, some patients will never get RT and will have less late effects but the remaining will need salvage therapy (more chemotherapy and use of RT) which would likely have more late toxicity than RT alone.

Reduction of RT volume by analysis of patterns of failure

In some tumors, careful patterns of failure analysis have led to a reduction of RT volume. A few decades ago, it was routine to deliver craniospinal irradiation in intracranial ependymoma. Pattern of failure analysis revealed that the primary site was the most common site of failure and <15 % of patients failed in the neuraxis [64, 65]. In intracranial germinoma, the whole ventricular field followed by a primary site boost can be treated in M0 disease when RT is used alone [16]. Previously, these patients have also been treated with craniospinal irradiation.

For medulloblastoma, there is now accumulating evidence that tumor bed alone can be treated instead of the entire posterior fossa during the RT boost portion of the treatment. Non-tumor bed posterior fossa failures are uncommon with current imaging and RT technology [13, 14].

Future direction

While many studies have tried to omit or delay RT, it has become clear that RT will have an important role in the management of many pediatric malignant brain and soft tissue tumors. Genetic testing may help identify subsets of patients who are more likely to develop radiation toxicity. In addition, future molecular and cytogenetic studies may reveal which tumors might be able to be cured without RT. For example, recent molecular subtyping of childhood medulloblastoma has revealed four distinct subtypes [66]. The WNT variety has been found to have the best prognosis and may be amenable to reduction of treatment. More studies are also needed to understand the pathogenesis of late effects of RT. For example, short-term memory loss has recently been linked to hippocampal irradiation. The periventricular and perigranular zones of the hippocampus, important sites for neurogenesis, are thought to be sensitive structures to radiation injury [67]. Several trials are underway using a hippocampal-sparing RT approach in the treatment of brain tumors.