Outcome and Toxicity of an Ifosfamide-Based Soft Tissue Sarcoma Treatment Protocol in Children. The Importance of Local Therapy

Background. Although the survival of children with soft tissue sarcoma (STS) has improved considerably, the outcome of patients with metastatic disease, and those with primary tumours of the extremities or parameningeal sites remains disappointing. We describe the clinical outcome of an ifosfamide-based regimen with local therapy directed only to children who failed to achieve a complete response to initial chemotherapy. Patients and Methods. Twenty-one children with STS (16 rhabdomyosarcoma) who presented with unresectable tumours were treated with five courses of ifosfamide (9 g/m2) and etoposide (600 mgm2). Patients who did not achieve a complete response then received local therapy. Chemotherapy with ifosfamide combined with etoposide, vincristine (1.5 mg/m2 and doxorubicin (60 mg/m2) or vincristine and actinomycin D (1.5 mg/m2) was continued for one year. Results and Discussion. Objective responses to five courses of ifosfamide and etoposide were seen in all patients. Disease free survival (DFS) at a median follow up of 59 months was 57% (95% CI 29–75%). The DFS of children who received local therapy was 89% compared with 33% in those who received chemotherapy alone (p=0.027). Locoregional recurrences did not occur in children who received radiotherapy to the site of the primary tumour. Ifosfamide-based chemotherapy does not reduce the incidence of loco-regional recurrence in children who do not receive local therapy.


Introduction
Soft tissue sarcom as (STS) constitute up to 6% of all childhood m alignancies. T he overall survival for children with rhabdom yosarcom a (RMS), the m ost com m on ST S, has increased considerably over the last tw enty-® ve years to 70% . 1 This im provem ent follows the introduction of m ultiagent chemotherapy com bined with effective treatm ent of residual tumour m asses with either surgery, radiotherapy (RT ), or a com bination of both. 2,3 T he outcom e of several subgroups of patients with RM S rem ains disap pointing. Features consistent with a poor prognosis include the presence of m etastatic disease, 3 high proliferative activity 4 and the presence of the [t(2;13)(q35;q14)] translocation. 5 Alveolar histology, previously thought of as being asso ciated w ith treatment resistance, m ay be of lesser signi® cance following the intensi® cation of chemotherapy regimens. 3 T he prim ary site of disease is also im portant with paratesticular and orbital tum ours having a m ore favou rable prognosis than those located in the param eningeal regions of the head and neck or the extrem ities. 2,3 The rarity of non-R M S ST S has led to uncertainty over the optimal treatment for these tum ours. Although generally considered to be less chemosensitive than RM S, m any are treated in a similar m anner with greater em phasis being placed on local disease control. 6 T wo ST S staging system s have been widely reported in the literature. T he post-operative clinical grouping strategy of the Intergroup Rhabdom yosarcom a Study Group (IRS) has been criticised as being strongly in¯uenced by the surgical app roach employed in each case. An alternative app roach has been adopted by the International Society of Paediatric O ncology (SIO P), w hich allocates treatm ent according to stage and post-surgical status. One consequence of the im proved survival of children w ith ST S has been an increasing awareness of the long term sequelae of treatm ent. These include the adverse effects of RT on growing bones, teeth, eyes, bladder and the pituitary gland, particularly in very young children. 7± 11 T he introduction of m ore intensive chem otherapy has been accom panied by the appearance of second m alignancies and m ay lead to a higher incidence of infertility. 12,13 D espite im proved survival the optim um com bination of chem otherapy and timing of local disease control rem ains uncertain. Here we report the results of a ifosfam ide-based treatment regimen in children who presented with tum ours which were not com pletely resectable at presentation without m utilating surgery.

P atient details
T wenty-one children (8 fem ale) with STS (16 RM S) w ere treated between N ovem ber 1987 and January 1994 using the treatm ent protocol described below (Table 1). T he m edian age at diagnosis was 4 years (range 0.6± 14 years). Eighteen children (85% ) presented with unresectable disease, either as a consequence of the site of the prim ary tum our or as the result of locoregional or distant spread. T hree patients (14% ) had m acroscopic/m icroscopic evidence of tum our following prim ary ). An identical translocation was also detected in an epithelioid schw annom a (No. 18). N o patient had received previous chem otherapy or RT .

Treatm ent protocol
All children initially received ® ve courses of Ifosfam ide (3 g/m 2 ) com bined with etoposide (200 m g/ m 2 ) on each of three consecutive days (IE). This com bination w as adm inistered for ® ve courses prior

Protocol violations
Both children w ith orbital RM S achieved a CR with ® ve courses of IE (Nos. 11 and 12). In view of this response, and the favou rable prognosis of this tum our site a further four courses of IE only were adm inistered. Another patient underw ent tum our resection after 10 courses of chemotherapy. N o viable tumour was found and therapy was discontinued in accordance with parental wishes (No. 17).

Local therapy
After com pleting ® ve courses of IE , 13 children (62% ) (nine RM S) had residual macroscopic disease ( Table 2). Six of these (four RMS) received RT at a m edian dose of 45 Grays (range 43± 50 G rays) to the tumour bed with m argins of at least 2 cm. T he patient with an intracranial extension of her param eningeal RM S received an additional 30 G rays w hole brain RT . Six children underw ent surgical resection of residual tum our.

Response
Re-evaluation following ® ve courses of chem otherapy demonstrated that all patients had objective responses to IE (see T able 2). T here were ® ve CRs (four RM S) (24% ) and 13 PRs (10 RM S) (62%). T he Kaplan-M eier estim ate of disease free survival (DFS) at a median follow up of 59 m onths (range 35± 115 m onths) was 57% (95% CI 29± 75% ). 16 D FS in children who received appropriate local therapy, either surgery or RT, was 89% at 3 years com pared to only 33% in those w ho received chem otherapy alone (p 5 0.027, Log-ran k test).
T wo children who developed recurrent disease without previously having received de® nitive local therapy were treated with further chemotherapy and RT . O ne is disease-fre e 17 m onths from recurrence (No 14). The second patient developed a further local recurrence 6 m onths later and is presently receiving palliative chem otherapy only (No 4).

D isease progression
In all there were ® ve locoregional recurrences (24% ). Four of these children w ere receiving treatm ent for RM S. A single patient developed pulm onary m etastases after com pleting treatment. T he patient with m ultiple bony m etastases from a RM S im proved during therapy but deteriorated upon its com pletion. O f the ® ve children w ho achieved a C R with chem otherapy none received de® nitive local therapy and tw o have subsequently developed disease recurrence at the site of the prim ary tum our. One patient who w as receiving treatment for a chest wall RM S accom panied by extensive m ediastinal and para-ao rtic lym phadenopathy achieved a to reassessm ent of the tumour. D uring these and subsequent courses Ifosfam ide was adm inistered as a continuous intravenous infusion over 72 hours accom panied by an equivalent dose of m esna included in the hydration¯uid which was given at 3 l/m 2 /day. M esna hydration w as continued for 12 hours after the infusion of ifosfam ide was com plete. Following ® ve courses of IE further chem otherapy consisted of rotating courses of Ifosfam ide (9 g/m 2 ) with vincristine (1.5 mg/m .3) and actinom ycin D (1.5 m g/m 2 (IV A), ifosfam ide w ith vincristine (1.5 m g/m 2 ) and doxorubicin (20 m g/m 2 ) adm inistered over 6 hours on each of three consecutive days (IV Ad) and further courses of IE (Fig. 1). C hem otherapy was adm inistered at three weekly intervals or delayed until the neutrophil count had recovered to 1 3 10 9 /l and the platelet count was greater than 100 3 10 9 /l. Total treatm ent duration was one year. Chemotherapy was continued during RT but courses of IVA or IVAd w ere replaced with IE, and the order of subsequent courses was rearranged to ensure that each child received the sam e cum ulative dose of each drug. In total, 21 children received a m edian of 15 courses of chem otherapy (range 9± 18 courses) comprising a m edian of 126 g/m 2 of ifosfam ide (range 81± 162 g/m 2 ) and a m edian of 240 m g/m 2 of doxorubicin (range 0± 240 m g/m 2 . Children with param eningeal RM S received eight doses of intrathecal chem otherapy com prising cytosine arab inoside (30 m g), hydrocortisone (30 m g) and m ethotrexate (12.5 m g). In all cases response w as evaluated by com puterised axial tomographic scans with enhancement.
Responses w ere de® ned as follows: Com plete Response (CR) resolution of all apparent tum our, Partial Response (PR) m ore than 50% decrease in tum our dim ensions as m easured by the sum of the products of the m axim al perpendicular diam eters of all m easurable lesions and Stable D isease less than 50% decrease in tumour dim ensions or no change.

Haematological and infective
A total of 293 courses of chem otherapy w ere adm inistered, 171 (58% ) of these w ere com plicated by an adm ission for fever in the presence of neutropenia. T here were 41 independent septicaemic episodes with a variety of infecting bacteria. Tw enty (49% ) of these proven infections were related to indw elling central venous catheter colonization. Two children developed episodes of m icrobiologically proven fungal septicaem ia. T here were two deaths from overwhelming sepsis. Com m on toxicity criteria grade 3 throm bocytopenia and grade 4 neutropenla were invariable following chem otherapy. H aem atological and infective toxicity was unaffe cted by concurrent RT but was m ore severe during the second 6 m onths of treatment. All children required interm ittent support w ith blood products.
Gastrointestinal C hemotherapy induced nausea and vomiting were m anageable w ith ondansetron com bined with dexam ethasone. W eight loss during therapy w as invariable and seven children (33% ) required nasogastric alim entation. C onstipation and anal ® ssures w ere com m on.

Renal
F ifteen children were available for renal function testing 2 years from diagnosis. T he median glom erular ® ltration rate was 76 m ils/min/1.73m 2 (range 59± 100 m ls/m in/1.73m 2 ) (see T able 2). A com bined score of both glom erular and tubular function in these children produced a m edian value of 5 (range 1± 10), with 4 to 7 representing m oderate nephrotoxicity and greater than or equal to 8 severe nephrotoxicity. 14 D etailed description of the extent of both glom erular and tubular dysfunction In m any of these patients has been presented elsewhere. 15 O ther toxicities W hilst increased som nolence during IF therapy was com m on, this rarely extended beyond 50% of waking hours (W HO G rade 2). Severe som nolence and confusion followed three courses of chemotherapy in three different patients (W H O Grade 3). In each case neurological im pairm ent resolved spontaneously and did not recur. Grade 3 neurotoxicity w as not considered an indication for adjusting subsequent chemotherapy. Local reactions to RT were uncom m on and no m ore severe than expected despite continuing chemotherapy with IE. C ardiac toxicity was m onitored using echocardiography and no patient has a persistent fractional shortening of less than 30% . Haem orrhagic cystitis requiring adm ission and platelet support occurred in two children. The patient who received cranial irradiation has persisting alopecia. F ollowing chemotherapy one patient developed a sym ptom atic oesophageal stricture as a consequence of severe m ucositis, and continues to require regular dilatation. Tw o children who received RT for a RM S of the m iddle ear have m arked unilateral hearing loss. T here have been no second m alignancies.

D iscussion
ST S of childhood represent a diverse group of tum ours which differ widely in term s of their m alignant potential and treatment response. RM S, the largest single m ember of this group, is increasingly recognised to be a spectrum of different diseases rather than a single entity. This is re¯ected in the variable prognosis for RM S at different anatom ical sites. 2,3 The location of the prim ary tum our does not provide a complete explanation as to w hy one child fares well with existing therapy, w hilst another patient with a histological and anatom ically identical tum our develops recurrent disease. In the future this issue may be clari® ed by m olecular studies. 17 T he presence of a chrom osom al translocation [t(2;13)(q35;q14)] in RM S has been linked w ith the presence of dissem inated disease at diagnosis and a poor response to treatment. 18 This translocation was detected in three patients, all of whom died from recurrence.
Here we report the use of an ifosfam ide-based treatment protocol in the m anagem ent of all non-resectable sarcom as. At its inception this program m e was designed to im prove cure rates In children w ho presented with tumours w hich w ere unresectable without m utilating surgery. Five courses of IE were given to shrink the tum our m ass prior to de® nitive local therapy. C hem otherapy w as continued for 1 year in an effort to eradicate systemic m icrom etastases. T he conclusions that can be drawn from these results are lim ited by both the sm all num ber of patients and the heterogeneity of the tum ours treated. T he Introduction of Ifosfam ide as the alkylating agent`backbone' of this regim en followed several prom ising reports of its ef® cacy In the treatm ent of STS and possible superiority over cyclophospham ide. 19± 26 An earlier report demonstrated the safe ty of 9 g/m 2 of ifosfam ide when administered over three days in children with norm al renal function. 27 Etoposide and ifosfam ide are synergistic in vitro and the combination has excellent activity against ST S in phase II studies. 28 T his activity was evident in the 100% response rate to ® ve courses of IE. T he sensitivity of STS to vincristine, actinom ycin D and doxorubicin is well established. 29± 31 Actinom ycin D and doxorubicin are known radiosensitisers and the adm inistration of either of these drugs during RT is accom panied by an increase in radiation-induced local tissue damage. 32 T his has led to both drugs being avoided during and im m ediately following RT. Although several studies have shown that etoposide inhibits the repair of sub-lethal radiation dam age in experim ental system s there is little evidence of an adverse reaction between epodophyllotoxins and RT in clinical practice. 33,34 As a consequence of this we chose to continue IE during RT rather than reduce the intensity of chemotherapy delivered. T his approach was not accom panied by an increase in the severity of local radiation toxicity. Although this ® nding requires con® rm ation in a larger study we suggest that the com bination of IE and RT , at the doses employed here, does not lead to prohibitive radiation toxicity.
O ur treatment regim en produced an overall D FS of 57% at a m edian follow up of 59 m onths. T he D F S of children who received de® nitive local therapy in this study was superior at 89%. This suggests that overall treatm ent ef® cacy m ay have been comprom ised by the om ission of local therapy from children who exhibited an excellent initial response to chem otherapy in an effort to reduce long-term sequelae. T he traditional view of de® nitive local therapy being necessary to eradicate ST S is being challenged in an effort to reduce the long-term sequelae of treatm ent. Current SIOP studies reserve surgery and RT for children with param eningeal RM S and patients who fail to achieve a CR with chem otherapy alone. Tw o of the ® ve children in this study who did not receive local therapy developed recurrent disease at the site of the prim ary tum our. O nly three children, two w ith orbital RM S, a particularly favou rable prognostic group, and one patient w ith retroperitoneal RM S rem ain disease free after chem otherapy alone.
The results presented here indicate that, despite intensifying the chem otherapy of STS, local therapy rem ains im portant in the prevention of locoregional recurrence. N o patient in this study who received de® nitive surgery or RT to the site of the prim ary tum our developed a local recurrence. T hese results highlight the dangers of withholding local therapy on the basis of im aging or repeat biopsy studies following an apparent C R to prim ary chem otherapy, except possib ly in the case of orbital RM S. T he decision to avoid RT in children with disease at other sites, especially elsewhere in the head and neck, following a C R to chem otherapy should be taken with caution and should be explored in a larger m ulticentre study. It has been suggested that decisions regarding further therapy are based upon the initial response to chem otherapy. 26 W e would advocate caution in this respect, particularly in making decisions about the necessity for surgery or RT. T wo out of ® ve children in this study who achieved a CR with chemotherapy, in the absence of local treatment subsequently developed a locoregional recurrence and died.
The toxicity of this regim en w as considerable. T wo children died from overwhelm ing sepsis (9% ), a proportion which is consistent with reports from other studies. 3,35 At the time of introduction of this treatment regim en ifosfam ide-nephrotoxicity was unknow n with early reports of ifosfam ide chemotherapy in children either failing to recognise renal toxicity or underestim ating its severity. Eight years later we recognise the im portance of this problem . 14 O verall the high incidence of long-term nephrotoxicity reported here re¯ects the large cum ulative dose of ifosfam ide adm inistered to m any of the patients. 15 In contrast to the relatively high adm inistered cum ulative dose of ifosfam ide, the total dose of doxorubicin w as only 240 m g/m 2 thus explaining the absence of signi® cant cardiotoxicity. A low inci-dence of signi® cant neurotoxicity was seen and is consistent with previous reports of the rarity of this condition in children. O ur approach of not altering subsequent ifosfam ide chemotherapy following encephalopathy was justi® ed by a zero recurrence rate.
This report demonstrates that despite intensive m ultiagent chem otherapy a high rate of loco-regional recurrence was seen am ongst children with STS who did not receive de® nitive local therapy. T his m ay have com prom ised overall survival. Patients who completed ifosfam ide chem otherapy with app ropriate local therapy did well but m any have signi® cant nephrotoxicity. T he initial response to IE was im pressive and w arrants comparison with other regim ens. W e do not advocate the w idespread introduction of this treatment protocol because of its considerable toxicity, in particular its nephrotoxicity. Instead, we believe this report illustrates the dif® culty in deciding which patients require local therapy and highlights the dangers of withdraw ing RT in favo ur of an Im pressive response to a new chem otherapy regim en. Further progress in optimising treatment regimens for STS will follow the com pletion of ongoing m ulticenter studies.