Childhood Cancer Survivors: Attained Adult Compared
with
Heights
Sibling Controls
Grace E. Holmes,* MD, Frederick F. Holmes,† MD, Ariel B. Baker,‡ BS, Ruth S. Hassanein,§ PhD
As part of a large collaborative study, the authors administered a pretested questionnaire to 62 childhood and adolescent cancer survivors and 62 same-sex sibling controls. The authors requested information regarding attained adult height on the questionnaire. Mean adult height of survivors (172.2 cm) was less than that of controls (174.1 cm), at a borderline significant difference of p 0.0757. Multivariate analysis examined four potential, independent variables as possible predictors of this difference. The presence of brain tumor (vs. nonbrain tumor) (p < .0001) and diagnosis at an early age (≤ 8 years vs. 9-15 years) (p .05) were factors significantly related to the differential; sex of patient and type of therapy were not. Our findings thus identify malignancy site and age at diagnosis as important predictors of adult height in childhood and adolescent cancer survivors. =
=
GROWTH
RETARDATION appears to be a of successful treatment of cancer in childhood and adolescence. Factors that may contribute to this include cranial irradiation with resultant growth hormone deficiency and irradiation of the spine with resultant epiphyseal damage. Chemotherapy, corticosteroids, the disease itself, infection, poor nutrition, and age of patient at diagnosis all have been postulated to affect ultimate height.1 A large collaborative effort with the National Cancer Institute by five institutions, the California Department of Health Services, the University of Iowa, the University of Kansas, the University of
complication
From the *Departments of Preventive Medicine and Pediatrics, the †Department of Medicine, the ‡Cancer Data Service, and the §Department of Biometry, University of Kansas Medical Center, Kansas City, Kansas. Supported in part by contract NCI-CP-01036, National Cancer Institute, National Institutes of Health. Correspondence to: Grace E. Holmes, Room 2016, Building 48, University of Kansas Medical Center, Kansas City, KS 66103. Received for publication July 1989, revised September 1989, and accepted October 1989.
Texas, and Yale
University was formed to study the long-term physical and psychosocial morbidity of childhood cancer survivors and their offspring as well as that of siblings of the survivors.’ Included in the study was information on attained adult height. The purpose of the present study was to compare attained adult height in long-time survivors of treated childhood and adolescent cancer with that of their same-sex adult sibling controls in the University of Kansas respondents of the above-mentioned study. A secondary purpose was to determine whether various factors were predictors of, or related to, any height differential that might be found to exist between cases and controls. We are unaware of any study on adult heights of childhood and adolescent cancer survivors in which same-sex healthy siblings were used as the control group. Materials and Methods
The Tumor Registry of the University of Kansas Medical Center (KUMC) has registered every cancer patient seen at KUMC since 1945, totalling 46,593
268
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
TABLE la. Tumors and Sites
50 to follow-up at a followof The whereabouts of all registered rate 99.9%. up childhood cancer survivors was known at the beginning of the study in 1981. Patients were selected who, during the period of 1945 through 1974, were 1) diagnosed at the University of Kansas Medical Center and treated for cancer between birth and 19 years of age; who had survived longer than 5 years; and who were older than 21 by December 31, 1979. The procedures of the study were in accordance with the ethical standards of the Human Subjects Committee at KUMC. For the purpose of this report, only those cases were used in which the cancer was diagnosed by the time the patient was 15 years old; the object was to avoid cases where the patient’s pubertal growth may
patients and losing only
have been nearly complete at diagnosis. Siblings of these patients served as controls, and up to two were chosen for participation by having their blood relationship, sex, and birthdate, in that order, matched to the index patient. Patients and controls were interviewed in person when possible; some interviews were conducted by telephone. When a patient or control had died or was unavailable, a relative was interviewed as a proxy. The interview instrument was a pretested questionnaire covering areas of education, occupation, habits, tobacco and drug use, general health, reproductive history, and health status of children. After a training period, the experienced interviewers were able to administer the entire questionnaire in about 1 hour. Height measurements were reported in feet and inches and later converted to metric units. Data were analyzed by paired t-test and by stepwise multiple regression by the use of the BMDP software package. The multiple regression was carried out with the dependent variable being the difference between the adult heights of the case-control pair. Potential, independent variables were 1) sex; 2) age at diagnosis; 3) presence or absence of brain tumor; and 4) treatment with radiation, chemotherapy, both, or surgery. These variables were studied to determine which might be significantly predictive of adult height differences between survivors and their samesex sibling controls. Results Of the 268
ling controls,
eligible 160
cases and their 369 sibfit the age of diagnosis (< 15 four of these had same-sex sib-
Kansas
cases
years) criterion. Sixty
ling controls. Two craniopharyngioma
by Sex far 62
Cases
excluded (one with with a pinealoma) because of the type of tumor and location. There were finally 62 pairs (40 male, 22 female) that could be matched with same-sex siblings for whom information on adult height was available. Tables 1 a and 1 b list tumors and sites by sex for the 62 cases. 7777-
cases were
and
one
-
-
$$j3oth male and temale adults with brain tumors ~ had significantly shorter adult height than their ~f the ~~;&dquo;same-sex sibling contr ’
T ?r±q±rmqym+ used.
The adult height differential between all cases (n = 62) and all controls was 1.9 cm, with the mean height of all cases being shorter (172.2 cm) than that of the controls (174.1 cm). This overall difference was of borderline statistical significance (p 0.0757, =
paired t-test). On multiple regression analysis the potential, independent variables were used to differentiate further the difference between the adult heights of the cases and controls. These were sex, diagnosis (brain tumor or no brain tumor), age at diagnosis (_s. no brain tumor.
270
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
at diagnosis and treatment was related to estimated loss in height. Malignancy itself has been cited as one factor related to skeletal growth, which is linked to the development of some malignant tumors. Fraumeni’ suggested a relationship between rapid skeletal growth and bone tumors in childhood and adolescence and Broomhall et al. found that children with untreated ALL, as a group, were significantly taller than the normal population.~ Hancock et a1.8 observed a possible tendency toward tallness in patients with Hodgkin’s disease.
lier age more
-
-
-
,
~~l~t~~~t~~i~ ~~~r~~~~~~~~~~~ ~~°ed~ctio~~ of adult An~ilie, and ..° ..... ’
%u Ci’x·x?’’= ...
FIG. 2. Mean adult heights of 62 childhood/adolescent cancer survivors (40 d, 22 S) and same-sex sibling controls, by younger or older diagnosis age.
Discussion
%
° ’
:.
_
_
,.
Conversely, Berry et al., in assessing height percentiles of 127 children with ALL, concluded that these children may have been short prior to therapy if diagnosed when they were younger than 4 years of TABLE 2. Mean Adult
Heights (cm) for Cases and Controls
It is generally appreciated that childhood cancer and its treatment may affect ultimate height of the patient, either directly or indirectly. The extent of long-term growth retardation in treated childhood and adolescent cancer survivors has been noted in several groups. Lippens, et al. ~ noted that cranial irradiation, but particularly craniospinal irradiation, had an adverse effect on height growth in children treated for acute lymphocytic leukemia (ALL) for varying time periods after diagnosis and treatment; Glayton et a/.~ concluded that chemotherapy was a more important growth-retarding factor. In studies of children treated for brain tumors not involving hypothalamic or pituitary areas, Albertsson-Wikland et al,4 found that growth deviation was related to the amount of radiation dosage, particularly when patients were irradiated at a younger age. Shalet et a1.5 noted that short stature occurred in patients with brain tumors treated with either cranial or craniospinal irradiation but that it was much more pronounced in the latter group. They concluded that growth hormone deficiency from either cranial or craniospinal irradiation could lead to a short spine but that skeletal disproportion was an additional insult produced by craniospinal irradiation. Again, ear-
271
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
age.’ Westphal et al. tients
at
noted that heights of ALL pashorter than those of a congroup of children with noncancerous, hema-
diagnosis
were
trasting tologic diseases.&dquo;
Pui et al. studied heights at initial diagnosis of malignancies in 3,657 children and adolescents younger
than 18 years and found no differences between the heights of children with malignancies and published standards for heights of children in control populations.ll Besshoi2 also found pretreatment heights of children with ALL no greater than those in matched controls with acute illnesses.
Regardless of height status prior to diagnosis of a malignancy, it is not surprising that patients treated for malignancies during their growing years do have effect, or both, on somatic standards for age and sex as height growth. well as healthy children have been used as controls.4,13 To our knowledge, this study is the first in which normal same-sex siblings have been used as controls and in which the closest sibling match possible short of identical twins was performed. Our results indicate that both male and female survivors of brain tumors had significantly shorter adult stature than their same-sex sibling controls, regardless of the treatment used. In this study the diagnosis of brain tumor turns out to have been the significant factor rather than any specific therapy used. After taking into consideration whether or not the survivors had a brain tumor, we found a significant association between age at diagnosis and height difference. Also, in survivors who did not have brain tumors, there was a tendency for controls to be taller if the cancer was diagnosed early. This was true for both sexes (Fig. 2 and Table 2). This relationship between poor growth and a young treatment age has been found by other investigators. 1,14
some
tumor or treatment
Normal
Our findings do not clarify the specific etiology of the short adult stature in childhood and adolescent cancer survivors. They do emphasize, however, that it is important for the clinician to consider both the site of malignancy as well as the age of diagnosis when discussing with patients and their families predictions of adult height.
References 1.
2.
Clayton PE, Morris-Jones PH, Shalet SM, et al. Growth in children treated for acute lymphoblastic leukaemia. Lancet 1988;27(1):460-2. Mulvihill JJ, Myers MH, Connelly RR, et al. Cancer in offspring of long-term survivors of childhood and adolescent cancer.
3.
Lancet
1987;ii:813-7.
Lippens RJ, Otten BJ, van’t Hof MA. Growth of children with acute lymphocytic leukemia: preliminary results. Hamatol Bluttransfus 1987;30:427-31.
4. Albertsson-Wikland K, Lannering B, Marky I, et al. A longitudinal study on growth and spontaneous growth hormone (GH) secretion in children with irradiated brain tumors. Acta Paediatr Scand 1987;76(6):966-73. 5. Shalet SM, Gibson B, Swindell R, et al. Effect of spinal irradiation on growth. Arch Dis Child 1987;62(5):461-4. 6. Fraumeni JF Jr. Stature and malignant tumors of bone in childhood and adolescence. Cancer 1967;20:967-73. 7. Broomhall J, May R, Lilleyman JS, et al. Height and lymphoblastic leukaemia. Arch Dis Child 1983;58(4):300-1. 8. Hancock BW, Mosely R, Coup AJ. Height and Hodgkin’s disease [Letter]. Lancet 1976;2(7999):1364. 9. Berry DH, Elders MJ, Christ W, et al. Growth in children with acute lymphocytic leukemia: a Pediatric Oncology Group study. Med Pediatr Oncol 1983;11(1):39-45. 10. Westphal M, Morgan SK, Grush OC. Nutrition and growth in children with acute lymphoblastic leukemia (ALL). Clin Res 1979;27:816A. 11. Pui CH, Dodge RK, George SL, et al. Height at diagnosis of malignancies. Arch Dis Child 1987;62(5):495-9. 12. Bessho F. Height at diagnosis in acute lymphocytic leukemia. Arch Dis Child 1986;61(3):296-8. 13. Griffin NK, Wadsworth J. Effect of treatment of malignant disease on growth in children. Arch Dis Child
1980;55(8):600-3. 14. Herber SM, Kay R, May R, et al. Growth of long term survivors of childhood malignancy. Acta Paediatr Scand
1985;74(3):438-41.
272
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
with
Heights
Sibling Controls
Grace E. Holmes,* MD, Frederick F. Holmes,† MD, Ariel B. Baker,‡ BS, Ruth S. Hassanein,§ PhD
As part of a large collaborative study, the authors administered a pretested questionnaire to 62 childhood and adolescent cancer survivors and 62 same-sex sibling controls. The authors requested information regarding attained adult height on the questionnaire. Mean adult height of survivors (172.2 cm) was less than that of controls (174.1 cm), at a borderline significant difference of p 0.0757. Multivariate analysis examined four potential, independent variables as possible predictors of this difference. The presence of brain tumor (vs. nonbrain tumor) (p < .0001) and diagnosis at an early age (≤ 8 years vs. 9-15 years) (p .05) were factors significantly related to the differential; sex of patient and type of therapy were not. Our findings thus identify malignancy site and age at diagnosis as important predictors of adult height in childhood and adolescent cancer survivors. =
=
GROWTH
RETARDATION appears to be a of successful treatment of cancer in childhood and adolescence. Factors that may contribute to this include cranial irradiation with resultant growth hormone deficiency and irradiation of the spine with resultant epiphyseal damage. Chemotherapy, corticosteroids, the disease itself, infection, poor nutrition, and age of patient at diagnosis all have been postulated to affect ultimate height.1 A large collaborative effort with the National Cancer Institute by five institutions, the California Department of Health Services, the University of Iowa, the University of Kansas, the University of
complication
From the *Departments of Preventive Medicine and Pediatrics, the †Department of Medicine, the ‡Cancer Data Service, and the §Department of Biometry, University of Kansas Medical Center, Kansas City, Kansas. Supported in part by contract NCI-CP-01036, National Cancer Institute, National Institutes of Health. Correspondence to: Grace E. Holmes, Room 2016, Building 48, University of Kansas Medical Center, Kansas City, KS 66103. Received for publication July 1989, revised September 1989, and accepted October 1989.
Texas, and Yale
University was formed to study the long-term physical and psychosocial morbidity of childhood cancer survivors and their offspring as well as that of siblings of the survivors.’ Included in the study was information on attained adult height. The purpose of the present study was to compare attained adult height in long-time survivors of treated childhood and adolescent cancer with that of their same-sex adult sibling controls in the University of Kansas respondents of the above-mentioned study. A secondary purpose was to determine whether various factors were predictors of, or related to, any height differential that might be found to exist between cases and controls. We are unaware of any study on adult heights of childhood and adolescent cancer survivors in which same-sex healthy siblings were used as the control group. Materials and Methods
The Tumor Registry of the University of Kansas Medical Center (KUMC) has registered every cancer patient seen at KUMC since 1945, totalling 46,593
268
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
TABLE la. Tumors and Sites
50 to follow-up at a followof The whereabouts of all registered rate 99.9%. up childhood cancer survivors was known at the beginning of the study in 1981. Patients were selected who, during the period of 1945 through 1974, were 1) diagnosed at the University of Kansas Medical Center and treated for cancer between birth and 19 years of age; who had survived longer than 5 years; and who were older than 21 by December 31, 1979. The procedures of the study were in accordance with the ethical standards of the Human Subjects Committee at KUMC. For the purpose of this report, only those cases were used in which the cancer was diagnosed by the time the patient was 15 years old; the object was to avoid cases where the patient’s pubertal growth may
patients and losing only
have been nearly complete at diagnosis. Siblings of these patients served as controls, and up to two were chosen for participation by having their blood relationship, sex, and birthdate, in that order, matched to the index patient. Patients and controls were interviewed in person when possible; some interviews were conducted by telephone. When a patient or control had died or was unavailable, a relative was interviewed as a proxy. The interview instrument was a pretested questionnaire covering areas of education, occupation, habits, tobacco and drug use, general health, reproductive history, and health status of children. After a training period, the experienced interviewers were able to administer the entire questionnaire in about 1 hour. Height measurements were reported in feet and inches and later converted to metric units. Data were analyzed by paired t-test and by stepwise multiple regression by the use of the BMDP software package. The multiple regression was carried out with the dependent variable being the difference between the adult heights of the case-control pair. Potential, independent variables were 1) sex; 2) age at diagnosis; 3) presence or absence of brain tumor; and 4) treatment with radiation, chemotherapy, both, or surgery. These variables were studied to determine which might be significantly predictive of adult height differences between survivors and their samesex sibling controls. Results Of the 268
ling controls,
eligible 160
cases and their 369 sibfit the age of diagnosis (< 15 four of these had same-sex sib-
Kansas
cases
years) criterion. Sixty
ling controls. Two craniopharyngioma
by Sex far 62
Cases
excluded (one with with a pinealoma) because of the type of tumor and location. There were finally 62 pairs (40 male, 22 female) that could be matched with same-sex siblings for whom information on adult height was available. Tables 1 a and 1 b list tumors and sites by sex for the 62 cases. 7777-
cases were
and
one
-
-
$$j3oth male and temale adults with brain tumors ~ had significantly shorter adult height than their ~f the ~~;&dquo;same-sex sibling contr ’
T ?r±q±rmqym+ used.
The adult height differential between all cases (n = 62) and all controls was 1.9 cm, with the mean height of all cases being shorter (172.2 cm) than that of the controls (174.1 cm). This overall difference was of borderline statistical significance (p 0.0757, =
paired t-test). On multiple regression analysis the potential, independent variables were used to differentiate further the difference between the adult heights of the cases and controls. These were sex, diagnosis (brain tumor or no brain tumor), age at diagnosis (_s. no brain tumor.
270
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
at diagnosis and treatment was related to estimated loss in height. Malignancy itself has been cited as one factor related to skeletal growth, which is linked to the development of some malignant tumors. Fraumeni’ suggested a relationship between rapid skeletal growth and bone tumors in childhood and adolescence and Broomhall et al. found that children with untreated ALL, as a group, were significantly taller than the normal population.~ Hancock et a1.8 observed a possible tendency toward tallness in patients with Hodgkin’s disease.
lier age more
-
-
-
,
~~l~t~~~t~~i~ ~~~r~~~~~~~~~~~ ~~°ed~ctio~~ of adult An~ilie, and ..° ..... ’
%u Ci’x·x?’’= ...
FIG. 2. Mean adult heights of 62 childhood/adolescent cancer survivors (40 d, 22 S) and same-sex sibling controls, by younger or older diagnosis age.
Discussion
%
° ’
:.
_
_
,.
Conversely, Berry et al., in assessing height percentiles of 127 children with ALL, concluded that these children may have been short prior to therapy if diagnosed when they were younger than 4 years of TABLE 2. Mean Adult
Heights (cm) for Cases and Controls
It is generally appreciated that childhood cancer and its treatment may affect ultimate height of the patient, either directly or indirectly. The extent of long-term growth retardation in treated childhood and adolescent cancer survivors has been noted in several groups. Lippens, et al. ~ noted that cranial irradiation, but particularly craniospinal irradiation, had an adverse effect on height growth in children treated for acute lymphocytic leukemia (ALL) for varying time periods after diagnosis and treatment; Glayton et a/.~ concluded that chemotherapy was a more important growth-retarding factor. In studies of children treated for brain tumors not involving hypothalamic or pituitary areas, Albertsson-Wikland et al,4 found that growth deviation was related to the amount of radiation dosage, particularly when patients were irradiated at a younger age. Shalet et a1.5 noted that short stature occurred in patients with brain tumors treated with either cranial or craniospinal irradiation but that it was much more pronounced in the latter group. They concluded that growth hormone deficiency from either cranial or craniospinal irradiation could lead to a short spine but that skeletal disproportion was an additional insult produced by craniospinal irradiation. Again, ear-
271
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
age.’ Westphal et al. tients
at
noted that heights of ALL pashorter than those of a congroup of children with noncancerous, hema-
diagnosis
were
trasting tologic diseases.&dquo;
Pui et al. studied heights at initial diagnosis of malignancies in 3,657 children and adolescents younger
than 18 years and found no differences between the heights of children with malignancies and published standards for heights of children in control populations.ll Besshoi2 also found pretreatment heights of children with ALL no greater than those in matched controls with acute illnesses.
Regardless of height status prior to diagnosis of a malignancy, it is not surprising that patients treated for malignancies during their growing years do have effect, or both, on somatic standards for age and sex as height growth. well as healthy children have been used as controls.4,13 To our knowledge, this study is the first in which normal same-sex siblings have been used as controls and in which the closest sibling match possible short of identical twins was performed. Our results indicate that both male and female survivors of brain tumors had significantly shorter adult stature than their same-sex sibling controls, regardless of the treatment used. In this study the diagnosis of brain tumor turns out to have been the significant factor rather than any specific therapy used. After taking into consideration whether or not the survivors had a brain tumor, we found a significant association between age at diagnosis and height difference. Also, in survivors who did not have brain tumors, there was a tendency for controls to be taller if the cancer was diagnosed early. This was true for both sexes (Fig. 2 and Table 2). This relationship between poor growth and a young treatment age has been found by other investigators. 1,14
some
tumor or treatment
Normal
Our findings do not clarify the specific etiology of the short adult stature in childhood and adolescent cancer survivors. They do emphasize, however, that it is important for the clinician to consider both the site of malignancy as well as the age of diagnosis when discussing with patients and their families predictions of adult height.
References 1.
2.
Clayton PE, Morris-Jones PH, Shalet SM, et al. Growth in children treated for acute lymphoblastic leukaemia. Lancet 1988;27(1):460-2. Mulvihill JJ, Myers MH, Connelly RR, et al. Cancer in offspring of long-term survivors of childhood and adolescent cancer.
3.
Lancet
1987;ii:813-7.
Lippens RJ, Otten BJ, van’t Hof MA. Growth of children with acute lymphocytic leukemia: preliminary results. Hamatol Bluttransfus 1987;30:427-31.
4. Albertsson-Wikland K, Lannering B, Marky I, et al. A longitudinal study on growth and spontaneous growth hormone (GH) secretion in children with irradiated brain tumors. Acta Paediatr Scand 1987;76(6):966-73. 5. Shalet SM, Gibson B, Swindell R, et al. Effect of spinal irradiation on growth. Arch Dis Child 1987;62(5):461-4. 6. Fraumeni JF Jr. Stature and malignant tumors of bone in childhood and adolescence. Cancer 1967;20:967-73. 7. Broomhall J, May R, Lilleyman JS, et al. Height and lymphoblastic leukaemia. Arch Dis Child 1983;58(4):300-1. 8. Hancock BW, Mosely R, Coup AJ. Height and Hodgkin’s disease [Letter]. Lancet 1976;2(7999):1364. 9. Berry DH, Elders MJ, Christ W, et al. Growth in children with acute lymphocytic leukemia: a Pediatric Oncology Group study. Med Pediatr Oncol 1983;11(1):39-45. 10. Westphal M, Morgan SK, Grush OC. Nutrition and growth in children with acute lymphoblastic leukemia (ALL). Clin Res 1979;27:816A. 11. Pui CH, Dodge RK, George SL, et al. Height at diagnosis of malignancies. Arch Dis Child 1987;62(5):495-9. 12. Bessho F. Height at diagnosis in acute lymphocytic leukemia. Arch Dis Child 1986;61(3):296-8. 13. Griffin NK, Wadsworth J. Effect of treatment of malignant disease on growth in children. Arch Dis Child
1980;55(8):600-3. 14. Herber SM, Kay R, May R, et al. Growth of long term survivors of childhood malignancy. Acta Paediatr Scand
1985;74(3):438-41.
272
Downloaded from cpj.sagepub.com at UNIV OF NEW HAMPSHIRE on March 11, 2015
