Long-Term Neurotoxicity of Chemotherapy in Adolescents and Young Adults Treated for Bone and Soft Tissue Sarcomas

Purpose. To study the long-term neurotoxicity of chemotherapy in adolescents and young adults treated for bone and soft tissue sarcomas. Patients and Methods. Thirty-six adolescents and young adults (median age 17 years) were examined following chemotherapy for bone and soft tissue sarcomas. Twenty-nine (29/36) had received cisplatin (median 400 mg/m2), 15/36 ifosfamide (median 20 g/m2), and 12/36 vincristine (median 16 mg). Neurotoxicity was assessed at a median of 8 months (range, 1–54 months) after completion of chemotherapy by clinical examination, nerve conduction studies, audiograms and autonomic function tests. The same nerve conduction studies were carried out in 20 normal volunteers to define normal ranges in this age group. Results. Sixteen patients (44%) had a significant reduction in deep tendon reflexes, and this clinical parameter correlated well with abnormalities detected in nerve conduction studies. Vibration perception threshold (VPT) was raised in 20/36 patients (55%) and this was the most sensitive single test in the assessment of neuropathy. There was a significant correlation between VPT and cumulative cisplatin dose received in mg/m-2 (r=0.607, p<0.01). Ten of 29 patients (35%) had abnormal nerve conduction studies with a pattern characteristic of sensory axonal neuropathy. No patient complained of auditory symptoms, but minor high tone hearing loss was detected by audiograms in 5/28 patients who had received cisplatin. No patients had symptoms of autonomic neuropathy, but autonomic function tests showed minor abnormalities in 4/22 patients tested, and all had received cisplatin. Conclusions. This study demonstrates significant, although asymptomatic, long-term neurotoxicity of cisplatin in adolescents and young adults receiving chemotherapy for bone and soft tissue sarcomas. Follow-up studies are planned to assess whether these neurological deficits improve with time.


Introduction
Intensive com bination chem otherapy has greatly improved the prognosis of childhood and adolescent tum ours. In osteosarcom as, for exam ple, the cure rates for UK children under the age of 16 years have risen from 20% in the m id-1970s, to 60% at the present time, an im provem ent attributable to intensive adjuvant chem otherapy. 1 Sim ilar advances are now taking place in the treatment of Ewing' s sarcom a. 2 T his is a considerable trium ph for m odern cancer management. The long-term adverse effects of cancer treatments in adolescents are of considerable im portance since, each year in the UK , 1200 young adults w ill have a history of having been cured of a cancer in childhood or adolescence. If their life span is norm al this w ill represent over 50 000 U K adults by the year 2030.
There is little published data on long-term neurotoxicity of chem otherapy in the adolescent age group. Indeed, the norm al ranges in adolescents are not well established for m any of the availab le tests of neurological function. Previous studies have not always stated what the reference population has been. 3,4 Although the reported neurotoxicity has not, in the m ain, been clinically severe, it is im portant nonetheless to establish the frequency and severity of sub-clinical neurotoxicity. Chem otherapy-in duced nerve damage may have im portant im plications for the future, because other potentially neurotoxic com pounds m ay have an additive neurotoxic effect in these patients. In later life, comm on illnesses such as diabetes m ay produce m ore severe neuropathy because the starting point is not one of a norm al peripheral nervous system .
The aim of the present study was to establish the frequency and severity of long-term sub-clinical neurotoxicity from cytotoxics currently used in adolescents and young adults with bone and soft tissue sarcom as, com pared with a reference population of norm al volunteers of sim ilar age, using the sam e m ethodology.

P atients and chem otherapy protocols
T hirty-six patients with bone and soft tissue sarcom as were studied. Patients' characteristics are shown in Table 1. T here were 19 m ales and 17 fem ales, with a m edian age at diagnosis of 17 years (range, 3± 29 years), and a m edian age at assessm ent of 19 years (range, 8± 30 years). Patients were assessed at a m edian of 8 m onths from the com pletion of chemotherapy (range, 1± 54 m onths). Tw entyone patients were between 1 and 12 m onths, eight patients between 13 and 24 m onths, and ® ve patients betw een 25 and 54 m onths from the completion of chem otherapy. D iagnoses w ere as follows: osteosarcom a (28), Ewing' s sarcom a of bone (® ve), soft tissue sarcom a (two), and spindle cell sarcom a of bone (one). All patients were treated on N ational and International protocols which included: (1) M RC osteosarcom a protocol BO 03, a random ised study of cisplatin/doxorubicin versus m ultidrug adjuvant chem otherapy; 5,6 (2) M RC/EOI (European O steosarcom a Intergroup) protocol for m etastatic or axial skeleton osteosarcom a; (3) U KCC SG (United K ingdom C hildren' s Cancer Study Group) protocol for Ew ing' s sarcom a of bone; (4) M RC pilot study in osteosarcom a of cisplatin, doxorubicin, ifosfam ide and high-dose m ethotrexate; (5) phase 2 study of m ethotrexate, ifosfam ide and doxorubicin in m alignant ® brous histiocytoma of bone. 7 T he protocols, chem otherapy and duration of treatm ent are show n in T able 2, and the actual doses of chem otherapy received are shown in Table 3.

C linical examination
All patients had full clinical neurological assessm ent and nerve conduction studies. T wenty-eight patients had bilateral audiogram s, and 22 patients had assessm ent of autonom ic function. Clinical neurological history included detailed questions concerning peripheral sensory and m otor neuropathy, unsteadiness, Lherm itte' s syndrom e, Raynaud' s phenomena. 8,9 auditory symptoms and sym ptom s of autonom ic neuropathy. Symptoms w ere assessed as either acute (i.e., im m ediate, during each chem otherapy course), intermediate (i.e., during the 6± 12 m onths of chem otherapy), or as late effects (after com pletion of all chem otherapy). Clinical neurological exam ination was carried out by the sam e investigator (HM E) w ith full neurological assessm ent of m otor and sensory system s, including joint position sense and vibration sense (tuning fork 128 c/s). D eep tendon re¯exes (DTRs) were assessed and scored as norm al (score 5 1), reduced (only present w ith reinforcem ent, score 5 0.5) or absent (score 5 0). Biceps, triceps, supinator, knee and ankle D TRs were scored to give a m axim um cum ulative norm al score of 10. Re¯ex scores of 6 or less were considered abnorm al.   end of the great toe so that the weight of the probe (300 gm ) w as supported by the toe. T he stim ulus w as increased from zero until the subject indicated having felt it, and the lowest of three readings was taken. 10 C utaneous thresholds for warm ing and cooling w ere recorded from the plantar surface of the foot using a standard portable system as described by Fow ler et al. 11 C riteria for overall abnorm al nerve conduction studies suggestive of a sensory axonal neuropathy were de® ned as follows: (1) absent sural SN AP (sensory nerve action potential); or (2) abnormal VPT w ith either one of the follow ing: (a) present but abnorm al sural SN AP; or (b) absent m edial plantar SN AP.

Autonom ic function tests
Sym pathetic function was measured by postural changes in blood pressure. T he subject lay supine for 20 m in before the BP was recorded at 1 m in intervals 3 3. On standing, BP was recorded every m inute for 5 m in and then at 10 m in. In norm al subjects the systolic blood pressure does not fall . 15 m mH g.
Parasym pathetic function w as assesse d by heart rate response to deep breathing and the Valsalva m anouevre. T he heart rate response to deep breathing was assessed using a continuous EC G during six full respiratory cycles over 1 m in. The V alsalva m aneouvre was used to evaluate parasym pathetic function by asking the patient to m aintain a forced expiratory pressure of 40 m m H g against an anaeroid pressure m onitor dial. 12 The heart rate ratio was assessed and recorded three times; a norm al ratio is The results were analysed using the values recom m ended by Ewing and Clarke. 12 A udiom etry H earing w as assessed by bilateral audiogram s and thresholds (dB) for detection were recorded between 250± 12 000 H z.

Statistical analysis
T he data were analysed using the Statview statistical package on an Apple Macintosh computer. T wo groups were com pared using Student' s t-test (continuous variables) and the chi-squ ared test (discrete variables). Signi® cant correlations with age were found for the SN AP am plitudes and the VPT s. Signi® cant correlation for height w as found for the f-w ave latency. Therefore, t-tests involving these variables were calculated after correction for age or height w here relevant. All correlations w ere determ ined by calculating Kendall' s rank correlation coef® cient (tau).

C linical parameters and audiograms
T he only clinical neurological abnorm ality reported w as Lherm itte' s sign, which was reported by one patient and occurred during chemotherapy. T here w ere no reports of symptomatic peripheral sensory or m otor neuropathy, sym ptom s due to autonom ic dysfu nction, Raynaud' s phenom enon, tinnitus or high tone hearing loss, as acute, interm ediate or delayed effects. C linical neurological assessm ent revealed no abnormalities of power (M RC scales), tone, or coordination, and no abnorm alities of the clinically assesse d sensory param eters, including joint position sense and vibration sense. However, the strikingly abnorm al clinical param eter was the cum ulative re¯ex score which was reduced to # 6 in 16/36 (44%) of patients. Bilateral audiogram s were abnorm al in 5/28 patients with m inim al loss of high tone hearing. All ® ve patients had received cisplatin (348± 607 m g/m 2 ; m edian, 473 m g/m 2 ).

A utonom ic function tests
Autonom ic function tests were mildly abnorm al in 4/22 patients tested. Tw o patients had m inimal sympathetic abnorm alities with systolic BP falling by 20 m m H g on standing. T here was evidence of m ild parasym pathetic abnorm ality in tw o patients with an abnorm al Valsalva ratio (1.2), and abnorm al response to deep breathing (heart rate increase , 15 beats/m in) in one patient. All four patients with m ild autonom ic dysfu nction had received cisplatin (200± 582 m g/m 2 ); three of these patients also had abnorm al V PT s, and two had abnorm al nerve conduction studies (NC S) suggestive of a sensory axonal neuropathy.

N europhysiology
N orm al lim its for the neurophysiological param eters were determ ined in healthy volunteers and expressed as m ean 6 2SD , except in the cases of m edial plantar and sural SN APs and VPT w here the m eans and SD s of the log-transform ed data were calculated and then converted back to original units to be used as lim its of norm al. 13,14 Results of all the neurophysio logical data are show n in Table 4. neuropathy correlated with the cum ulative total dose of cisplatin (m g/m 2 ) and re¯ex score, but showed no correlation with the treatm ent-free interval, or the dose of vincristine received (Table 5). N one of the seven patients who received less than 300 m g/m 2 of cisplatin had neurophysiological evidence of sensory axonal neuropathy. T here was also a signi® cant correlation between re¯ex score and total received cisplatin dose (r 5 2 0.483, p 5 0.009).

C orrelation of re¯ex score with N CS abnorma lities
T he only consistently abnorm al clinical param eter in this group of adolescents and young adults was the re¯ex score. H owever this did prove useful as a sim ple clinical m arker of an otherw ise sub-clinical neuropathy. There was a signi® cant correlation between re¯ex score and VPT (Fig. 2, r 5 2 0.548, p , 0.001), m edial plantar and sural SN AP, and m edial plantar SCV (Table 6). Re¯ex score (norm al or abnorm al) correlated with both abnorm al N C S (p , 0.005, Table 5) and abnorm al V PT (p , 0.005).

D iscussion
C isplatin is w idely used in the treatment of bone sarcom as, but does produce well-recognised neurotoxicity which can lim it its use. 15± 17 Factors in¯uencing the degree of neurotoxicity from cisplatin include total cum ulative dose 15 are relatively poorly protected by the blood barrier and, as such, seem to be the m ost vulnerable neural structures to the effects of neurotoxins. 30,31 M arked degeneration and gliosis of the dorsal colum ns with signi® cant axonal loss of the dorsal roots were reported in a 3-year-o ld girl w ho had received 960 m g/m 2 of cisplatin. 32 The dorsal column involvem ent is thought to be responsible for the occasional case of Lherm itte' s sign reported in association with the neuropathy. 33,34 There is still som e uncertainty whether the m ajor cisplatininduced dam age to the hum an sensory peripheral nerves is at the level of the cell body, the axon, or both. Although cisplatin toxicity is know n to involve predom inantly large-diam eter sensory ® bres, an autonom ic neuropathy has been described 14,35 occurring in patients receiving a com bination of cisplatin, vinblastine and bleom ycin, and this has been postulated to be a`dying back' sm all-® bre neuropathy. A m otor com ponent to the neuropathy has been reported to occur in som e patients receiving high doses of cisplatin. 36 This report adds to the experience of cisplatin neuropathy in young people, and provides a comparison with an age-m atched control population. W e have found evidence of a persisting sensory axonal neuropathy in this group of young people who received cisplatin as part of adjuvant chem otherapy treatment for bone and soft tissue sarcom as. T en of 29 patients (33% ) w ho received cisplatin had evidence of a sensory axonal neuropathy at a m edian of 8 m onths after the com pletion of chem otherapy. T his is m anifest by a reduction in the am plitude of the sensory action potentials in sural and m edial 300 m g/m 2 or greater, and at doses greater than 600 m g/m 2 severe disabling sensory ataxia and m otor problems m ay be seen. Increases in VPT are often the earliest m anifestation of peripheral neuropathy 25 usually preceding sym ptom s and signs by 6± 8 weeks. 26 T hey are found to be signi® cantly elevated in patients who have received doses of cisplatin . 300 m g/m 2 . 27 T his is followed by a reduction in am plitude 4 or absence 28 of sensory nerve action potentials, consistent with a sensory axonal neuropathy or ganglioneuropathy. It appears that the neuropathy is due to a toxic effect on the dorsal root ganglia, where tissue platinum levels have been found to be the highest. 18 37 is uncertain, but seems unlikely. O f som e interest is the ® nding of signi® cant prolongation of f-w ave latencies in the patient group as a whole, w hich persisted after controlling for height, and the relative slowing in m otor conduction in the posterior tibial nerve. As there is no sensory com ponent to the f-w ave latency m easurem ent, this suggests that subtle m otor involvem ent is detectable, even in m ild asym ptom atic cases such as those reported here. T his could be a result of either a m ild diffu se disorder of m yelination causing a slight slowing of conduction velocity, or it could be the result of m ild axonal loss in which the large fastest conducting ® bres are lost ® rst, resulting in a change in the m inim um f-w ave latency, but not suf® cient to cause a noticeable decrease in the CM AP am plitude. T he degree of axonal loss could be assessed by further EM G studies of such a group of patients.
The V PT emerged in this study as the m ost valuable single neurophysiological test in the assessm ent of cisplatin-induced neuropathy in our patients.
M easurem ent of V PT using the Biothesiom eter is a sim ple technique which does not require the experienced training necessary for the carrying out of other m ore sophisticated nerve conduction studies. The Biothesiom eter is cheap, and easily portable, and the test itself takes less than a m inute to perform and is entirely painless for the patient. It is a quantitative m easure which signi® cantly correlated with cum ulative cisplatin dose, and could also be of use particularly in following recovery after com pletion of chemotherapy. In further research into cisplatin peripheral neuropathy, VPT m easurem ent using a Biothesiom eter could be recom m ended. This will be particularly useful for long-term follow -up of younger patients. D evelopm ents in this area of clinical cancer research are likely to include the developm ent of neuropro-tectors, and the ease and acceptability of V PT m easurem ent with the Biothesiom eter would m ake this an obvious choice for collection of repeated m easurem ents within the context of randomised clinical trials.
Reports concerning the progress of the neuropathy are anecdotal. Improvem ent of the neuropathy over time has been reported by som e authors but not by others. 25 It seems that som e im provem ent m ay occur after cessation of therapy, but de® cits do rem ain affecting vibratory perception and proprioception, which are considered largely irreversible 38 and which m ay progress. 39± 41 In our study there was no correlation between the time from chem otherapy and any of the clinical or neurophysiolo gical param eters, and this would suggest that there is not signi® cant recovery, through time, of the subclinical peripheral neuropathy documented in these young people.
Cisplatin is used in combination with other cytotoxics in a num ber of chemo-curable m alignancies which affect young people, including bone sarcom as and germ cell tum ours. There has been recent concern that its substitution by the less neurotoxic analogue carboplatin in the treatment of m etastatic germ cell tum ours has led to a signi® cant worsening in outcom e, because of a possible reduction in ef® cacy. 42 It is likely that, for the foreseeable future, cisplatin w ill rem ain an im portant com ponent of com bination chem otherapy in a num ber of m alignancies in adolescents and young adults. O ur study adds further to the evidence of substantial neurotoxicity of cisplatin in this group, in com parison with age-m atched controls, and lends im portant evidence to argum ents for the developm ent of effective neuroprotectors. 21,43 The precise m olecular m echanism s of the neurological dam age produced by cisplatin are unknown. Possibilities include the reaction of cisplatin with D N A in the dorsal root ganglia to reduce the production of nerve growth factor, or other neurotrophic proteins responsible for the m aintenance of an intact peripheral nervous system . O ther possibilities include the reaction of cisplatin with neurotransport proteins. W hatever the m echanism s of toxicity, the developm ent of neuroprotectors will be increasingly im portant. Several agents have already been developed and used in this role. 44 T hese include W R 2721, a thiol derivative, which has been used with the aim of reducing haem atological, renal and neurotoxicity. 39 O rg 2766, a fouram ino acid derivative of AC TH 45 has been used successfully in the prevention and am elioration of neurotoxicity in patients with ovarian cancer. 46 Exciting developm ents have recently been reported, with the newly recognised nerve growth factors in the regeneration of the peripheral nervous system 47 and the prevention of experim ental cisplatin neuropathy. 48 Insulin-like growth factor-I (IG F-I) 49 has also been found to be a potent neurotrophic agent in in vivo testing in an experim ental rat m odel and is entering Phase 2 testing in degenerative neurological diseases. However IG F-I' s role as an active grow th factor in m any tum ours has delayed its testing in cancer patients until the results of further in vitro and in vivo tests are available.