Longitudinal Growth Following Treatment for Osteosarcoma

Purpose. The purpose of this study was to analyse the height at diagnosis and growth in 72 skeletally immature children who had been treated for osteosarcoma in the area of the knee. Subjects. Of the patients, the average age at diagnosis was 10 years in girls and 12 years in boys. All children received neo-adjuvant chemotherapy, and had limb salvage by endoprosthetic replacement. Results and conclusion. The results of this study indicate that there is no evidence that children with osteosarcoma are taller at diagnosis than their normal counterparts. However, there was a marked retardation in growth in the year following the administration of cytotoxic chemotherapy. There were 19 children who reached skeletal maturity. The final height in these children was not significantly different from the normal population.


Introduction
It has been reported that children with osteosarcom a are taller than their normal counterparts. 1 However, this has been contradicted by others. 2,3 Glasser et al. 2 show ed that children with m alignant prim ary bone tum ours had a m arked retardation in growth during the year of cytotoxic chem otherapy. T hey also suggested that their ® nal height might be affe cted. It was the purpose of this study to analyse the height at diagnosis and the grow th that had occurred in skeletally im mature children with an osteosarcom a around the knee who have been treated by lim b salvage at our centre.

S ubjects and m ethods
All skeletally im m ature children who were diagnosed between 1981 and 1994 as having an osteosarcom a in the area of the knee, that had been treated at our centre by lim b-salva ge surgery, were included in this study. Patients were excluded if they died within 1.5 years of diagnosis, or if height m easurem ents were unavailable.
At diagnosis, all patients were fully staged and had their bone age estimated according to the G reulich and Pyle 4 m ethod. H eight was also m easured in the patients, w hile standing with a level pelvis. 5 Follow ing diagnosis, all patients received neoadjuvant chemotherapy according to the then current protocol. T his w as norm ally cisplatinum and doxorubicin or cisplatinum , doxorubicin and highdose methotrexate. 6 After three cycles of chem otherapy (usually 9 weeks following diagnosis), patients were restaged and underwent lim b-salva ge surgery. Lim b salvage was perform ed by resection of the tum our and endoprosthetic replacement. T he expected growth in the resected segm ent was calculated with the aid of the bone age and data provided by T upm an. 7 If the expected growth in the resected segm ent exceeded 3 cm , an extensible endoprosthetic replacement was inserted, 8± 10 otherw ise a non-extensible replacement that allowed some norm al growth to continue was used. 10± 12 Patients were discharged after 2 weeks and had a further three cycles of chem otherapy. All children were followed-up in the outpatient clinic and had their height estimated at regular intervals until they reached skeletal m aturity. C hildren who had an extensible prosthesis had regular lengthenings in order to m aintain lim b length equality. On average, two lengthening operations per year, from time of diagnosis until they reached skeletal m aturity, were perform ed. 13 The heights at diagnosis of all children were plotted in the growth charts for British children 14,15 according to guidelines given by Cole. 16 The children were split into three groups. One group of patients below the 25th percentile, one group between the 25th and 75th percentiles and one group above the 75th percentile. Statistical analysis was perform ed with the aid of the Chisquare test.
The heights of the children 1 and 5 years following diagnosis, and at skeletal m aturity, w ere also plotted in the growth charts. Statistical analysis was perform ed in a m anner identical to that described above.
The standard deviation score (or Z score) is the actual height of the patient m inus the m ean height of the population for that chronological age divided by the approp riate standard deviation. T herefore, a standard deviation score of 0 indicates that the patient has an average height for age and gender. Sim ilarly, a child with a score of 1 1, is one standard deviation above average for age and gender. T he difference in standard deviation score at diagnosis and 1 year following treatm ent is a m easure for the velocity of growth. A difference in standard deviation score of 0 indicates that the child had an average velocity of growth for that age range.
The standard deviation score was calculated for all patients who had their height estimated at diagnosis and after 1 year following treatm ent. T he difference in standard deviation score after 1 year w as calculated (negative is a decrease in standard deviation score). Sim ilarly, the difference in standard deviation score was calculated 5 years following treatment and at skeletal maturity.
All patients had their bone age estimated at time of diagnosis. T he bone age w as estim ated according to the m ethod described by Greulich and Pyle 4 and com pared to chronological age.

Results
Seventy-tw o children were identi® ed from the m edical records. T here w ere 47 boys and 25 girls. T he average age at diagnosis w as 12.1 years (range 5.8± 15.3) in boys and 9.9 years (range 5.8± 13.5) in girls. In 35 cases the left leg was affe cted and in 37 cases the right leg. There w ere 45 children w ho had a distal fem oral, and 27 who had a proximal tibial, osteosarcom a. Lim b salvage was by an endoprosthetic replacement that could be lengthened in 50 children. The rem aining 22 children had an endoprosthetic replacem ent that could not be lengthened, but that allowed for som e growth to continue. 10± 12 All children received neo-adjuvant chem otherapy. F orty-three children received cisplatinum and doxorubicin only, and 17 children received cisplatinum F ig. 1. D ifference in bone age and chronological age at diagnosis in 72 children with an osteosa rcom a in the area of the knee. A nega tive value indicates that the bone age was less than the chronologica l age. and doxorubicin in com bination with high-dose m ethotrexate. 6 Of the rem aining 12 children, nine had chemotherapy according to the regime proposed by Rosen et al. 17 and three children had alternative treatment protocols. There were 22 children who eventually died. T wenty deaths w ere due to m etastatic disease, one due to doxorubicin-induced cardiom yopathy, and one patient died due to extensive ileofemoral thrombosis. All but one patient died within 5 years of diagnosis, but no patient died prior to 1.5 years following diagnosis.
The bone age at diagnosis was on average 9.7 years (range 6± 13) in girls and 11.3 years (range 5± 14) in boys. As can be seen in Fig. 1, the vast m ajority of children had a bone age that was less than the chronological age (on average 0.25 years in girls and 0.73 years in boys). This difference was statistically signi® cant (sign test, p , 0.05).
There w ere 63 patients who had their height estim ated at diagnosis. T he results are shown in Fig.  2. There w as no statistically signi® cant difference  14,15 between the children in our study and the population norm . 14,15 In 50 children, the height was m easured 1 year following diagnosis (Fig 3). T he heights in these children were also not signi® cantly different from the norm al population. 14,15 Forty-one of these 50 children also had their height estimated at diagnosis. As can be seen in Fig. 4, the vast m ajority of children had a standard deviation score that was less than the standard deviation score at diagnosis. In girls, the standard deviation score 1 year following diagnosis was, on average, 0.67 less than that at diagnosis, w hilst this was 0.30 less in boys. T his difference w as statistically signi® cant (sign test, p , 0.05), indicating that chem otherapy causes retardation in growth in the year following treatment.
At 5 years following diagnosis, 18 children had their height m easured. T he heights in these children w ere also not signi® cantly different from the normal population. 14,15 Fifteen of these children also had their height measured at diagnosis. In girls, the standard deviation score was, on average, 0.17 less than the standard deviation score at diagnosis. Boys had a standard deviation score that w as on average 0.25 less than that at diagnosis.
There were 19 children in our study w ho reached skeletal m aturity. The ® nal height in these children was not signi® cantly different from the norm al population (Fig 5). Sixteen of these children also had their height estim ated at diagnosis. The standard deviation score in girls who reached skeletal m aturity was, on average, 0.48 higher than that at diagnosis, whilst this was 0.24 higher in boys.
These results indicate that the surviving children subsequently m ake up for the retardation in growth caused by the chem otherapeutic treatment. Although the num ber of children who reached skeletal m aturity is relatively sm all, ® nal height does not seem to be different from the population norm . 14,15

D iscussion
T his study failed to reproduce the earlier ® ndings of Fraum eni, 1 suggesting that children with osteosar-F ig. 4. Difference in standard deviation score after 1 year in 41 children who had their height estim ated at diagnosis and after 1 year. A negativ e value indicates that the child had a less than average grow th for age and gend er.
F ig. 5. Height at skeleta l maturity in 19 children who have been treated for an osteosarcom a in the area of the knee plotted in the grow th charts for the norm al population. 14,15 adm inistration of cytotoxic chemotherapy. T his further illustrates that actual height is a poor indicator of growth retardation and that the velocity of growth is a m ore sensitive m easure. The ® nal height of the children in this study was sim ilar to the norm al population. 14,15 Furtherm ore, the difference in standard deviation score was, on average, positive. T his indicates that these children subsequently m ake up for the retardation in growth caused by the chemotherapy treatment.
In conclusion, there is no evidence that children with osteosarcom a are taller than their norm al counterparts. Grow th is retarded in the year follow ing chem otherapy treatment, but the ® nal height in these children is not signi® cantly different from the norm al population. O steogenic sarcom a is m ore com m on in the second decade of life, during increased skeletal growth. It seems therefore likely that the incidence of osteosarcom a is related to the velocity of grow th at time of diagnosis rather than the height at diagnosis. U nfortunately, no pre-diagnostic height m easurem ents were available from the children in this study. T his makes it dif® cult to draw any conclusions. How ever, it is interesting to note that the vast m ajority of children had a bone age at diagnosis that w as less than their chronological age. It m ight be that these children were catching up on their relative skeletal imm aturity and had an increased velocity of growth at time of diagnosis.
The height 1 year follow ing chemotherapy treatm ent was also not signi® cantly different from the population norm . 14,15 H owever, there was a signi® cant reduction in standard deviation score, indicating that growth is retarded following the