D.B. Allen
`
`Limitations of short- term
`studies in predicting long-term
`adverse effects of inhaled
`corticosteroids
`
`Authors' affiliations:
`D.B. Allen, Department of Pediatrics, University
`of Wisconsin School of Medicine, Madison, M,
`USA
`
`Correspondence to:
`David B. Allen, MD
`H4/448 CSC - Pediatrics
`University of Wisconsin Children's Hospital
`600 Highland Avenue
`Madison, M 53792
`USA
`Tel. 608-263-5835
`Fax: 608-263-0440
`E-mail: dballen@facstaff.wisc.edu
`
`To cite this article:
`Allen D.E Limitations of short-term studies in predicting
`long-term adverse effects of inhaled corticosteroids.
`Allergy 1999, 54( 29-34.
`Copynght 0 Munksgaard 1999
`
`Key words: adrenal suppression; adverse effects; bone mineral
`density; growth; inhaled corticosteroids.
`
`This paper examines the value of short-term studies in predicting
`long-term, clinically relevant adverse effects of inhaled cortico-
`steroids (ICS) in children with asthma. Increasing use of ICS in
`younger and less severely affected children with asthma justifies
`concern about
`long-term adverse effects. For each system
`potentially affected by ICS, short-term and sensitive studies have
`limited value in predicting clinically relevant effects, even when
`correlations are highly statistically significant. This is due t o
`inherent limitations of the short-term tests utilized and normal
`physiologic variations in systems being studied. Specific limitations
`include:
`
`1) poor distinction between systemic presence of ICS and adverse
`systemic effects (e.g., integrated plasma cortisol)
`2) lack of data validating the connection between test results and
`the end point in question (bone markers t o predict growth and
`fracture risk)
`3) sensitivity confounded by normal physiologic variation (knemo-
`metry t o predict long-term growth).
`
`Consequently, predicting clinically relevant long-term effects from
`short-term studies detecting physiologic perturbations remains a
`challenge. Positive predictive value is improved by well-designed
`intermediate-term
`(>12 months) studies utilizing dynamic
`hypothalamic-pituitary-adrenal (HPA) axis testing, dual x-ray absorp-
`tiometry (DXA) scanning, or precise stadiometry. Ultimately, how-
`ever, long-term studies are required t o assess long-term risks, and the
`reliabilitv of short-term assessments in oredictina them.
`
`EX2200
`Eli Lilly & Co. v. Teva Pharms. Int'l GMBH
`IPR2018-01427
`
`1
`
`

`

`Allcn . Short-term studics of ICS
`
`Inhaled corticosteroids [ICS) are being prescribed for younger
`children, with greater consistency of administration, and for
`longer periods of time [ 1 ) . Appropriately, concern about
`potential long-term adverse effects of ICS treatment persists.
`Numerous studies document the effects of systemically
`absorbed ICS on the hypothalamic-pituitary-adrenal [HPAJ
`axis, bone metabolism, and growth. Not surprisingly, there
`continues to be hope that the findings of short-term studies,
`which are relatively easy to conduct and to control, will
`predict long-term adverse effects of ICS.
`the techniques
`Unfortunately, the high sensitivity of
`employed in short-term studies often compromises positive
`predictive value for the adverse effect in question. Thus,
`while short-term studies effectively identify systemic
`cffects of ICS, and even shed light on underlying mechan-
`isms, they fail to predict accurately the long-term con-
`sequences of these effects. Increasing duration of study leads
`to improved positive predictive value for virtually all the
`clinically relevant adverse effects of ICS. A role for short-
`term tests of ICS systemic effects to screen for individuals
`requiring more intensive safety monitoring remains to be
`defined.
`
`Test characteristics
`
`The validity of a test to detect the presence or absence of an
`adverse drug effect can be judged by its sensitivity and
`specificity. Sensitivity, the true positive rate, is the
`probability that a person demonstrating an adverse effect
`will have a positive test result. Sensitivity approaching
`100% avoids underdiagnosis, but also results in greater false-
`positive rates. Specificity, the true negative rate, is the
`probability that a person not demonstrating the adverse
`effect will have a negative test result. Of greater value to the
`clinician is the positive predictive value, which, for
`purposes of our discussion, indicates the proportion of
`individuals who ultimately manifest the clinically relevant
`adverse effect after a positive short-term test. The lower the
`positive predictive value, the greater the number of patients
`mistakenly identified as at risk of an adverse effect.
`Optimally, short-term tests of ICS effects would have high
`sensitivity for detecting changes known to lead
`to a
`clinically relevant adverse effect, accompanied by a high
`positive predictive value to minimize false-positive predic-
`tions of treatment complications and unnecessary disconti-
`nuation of effective therapy. Fulfillment of these criteria by
`tests and techniques commonly used in short-term studies
`of ICS is diminished by the following:
`
`1) poor distinction between systemic presence of ICS and
`clinically important systemic effects (e.g., integrated plasma
`cortisol)
`2) lack of data connecting test result to the end point in
`question (bone markers to predict deficient mineral accre-
`tion or fracture risk)
`3 ) sensitivity confounded by normal physiologic variation
`(knemometry to predict long-term growth).
`
`Systemic drug presence vs adverse effect:
`H P A axis
`
`The challenge of distinguishing detectable ICS-induced
`physiologic perturbations from clinically important adverse
`effects is particularly relevant to the HPA axis. Complica-
`tions of glucocorticoid (GC} therapy occur when total GC
`exposure is greater than normal and/or the pattern of
`exposure is significantly altered from normal diurnal
`variation. Consequently, replacement of endogenous GC
`effects by systemically absorbed IC does not necessarily
`mean that effects of GC excess will occur.
`Various tests of adrenal function reflect distinct aspects of
`HPA axis integrity [Table 1 ) (2). Urinary free cortisol
`excretion or area-under-the curve [AUC) 1%- or 24-h plasma
`cortisol concentrations are very sensitive measures of
`endogenous cortisol production, which can be proportio-
`nately decreased by the presence of any systemically
`absorbed ICS. Consequently, statistically significant sup-
`pression of endogenous cortisol production is necessary, but
`frequently not sufficient, for development of clinically
`important HPA axis suppression. After bedtime adminis-
`tration of 200 pg of either beclomethasone dipropionate
`[BDP) or budesonide [BUD) to children with asthma (3), the
`nocturnal rise in cortisol is delayed [reflecting blunting of
`the early nocturnal ACTH surge) and AUC cortisol secretion
`reduced. On the other hand, after temporary suppression,
`cortisol levels rise to normal morning levels, indicating
`preservation of the diurnal rhythm of the HPA axis. Thus,
`while systemic presence of IC is demonstrated by AUC
`analysis, the predictive value of this finding for clinically
`important HPA axis suppression is likely to be low (Fig. I).
`Short-term studies of HPA axis function do allow prompt
`comparisons between the systemic “cortisol substitution”
`effects of various ICS (4). Reflecting differences in receptor
`binding affinity and other pharmacokinetic attributes, more
`potent medications (e.g., fluticasone dipropionate [FP]) show
`greater suppression of endogenous cortisol than less potent
`
`2
`
`

`

`I
`
`AUC cortisol
`1 UFC
`
`I
`
`ACTH stimulation
`
`t_
`I (law or high dose)
`’
`Insulin-induced
`I hypoglycemia ~n~~~~~~
`1
`Of
`insufficiency
`Positive predictive value for important adverse effect
`
`---
`-
`-
`- _ _ _ _ ~ _ _ _
`Sensitivity to detect systemic IC presence
`
`~
`
`I
`
`Allen , Short-term studies of ICS
`
`by stad~ometry) associated with the use of ICS. Growth
`suppression (measured by knemometry) in children receiving
`400-800 pg/day BDP was associated with reductions in markers
`of collagen degradation skin and loose connective-tissue
`collagen synthesis, but not in the major collagen in bone (i.e.,
`type 1 procollagen) (7). While evidence supports IC-mediated
`effects on collagen metabolism as a mechanism for growth
`suppression (8), the value of biochemical markers in predicting
`growth remains limited even a highly statistically significant
`correlation (e.g., 0.5) indicates that only a low percentage of the
`vanation in growth may be explained by the marker.
`Most of the information on the effects of ICS on bone
`turnover has been obtained in adults treated with moderate-
`to-high doses, Measurements of BMD in children treated
`with BDP Or BUD have been
`to those in untreated
`children (13). Markers of bone resorption (e.g., urinary
`deoxypyridinoline) are unaffected by moderate-dose BDP
`treatment, even in children who exhibit growth suppression
`(14). On the other hand, serum osteocalcin levels can be
`reduced in BDP-treated children with normal BMD, suggest-
`ing that asthma per se might reduce osteocalcin indepen-
`dently of corticosteroid effect (13).
`Since normal BMD and normal final height have (thus far)
`been characteristic of children treated with ICS (9, 10, 151,
`there is, as yet, no valid association between short-term bone
`metabolism markers and a clinically important end point.
`Until there is documentation of the long-term effects of ICS on
`BMD, changes, for instance, in biochemical markers are likely
`to overestimate clinical risk. Studies of elderly patients have
`only recently identified markers (e.g., urinary type 1 collagen
`N-telopeptide) which efficiently distinguish between normal,
`osteopenic, and osteoporotic BMD already present (16). No
`studles have shown prospectively that bone markers evaluated
`early in the course of IC treatment predlct subsequent
`normality or adverse changes in BMD. Even dlrect measure-
`ment of BMD by either quantitative microdensitometry or
`dual x-ray absorptiometry (DXA), while capable of predicting
`risk of fracture, cannot reliably identlry indlvidual people who
`will have a fracture (17). In summary, short-term markers of
`bone turnover sensitively identlfy and even provide insights
`into mechanisms of the effect of ICS on bone, but have not
`been shown to predict consequences of this effect (Fig. 2).
`
`Confounding by normal physiologic
`variations: knemometry
`
`Childhood growth is complicated by changing factors
`regulating growth, normal variation in the tempo of long-
`
`Allergy 54, / 29-34
`
`I 3 1
`
`Figure 1 . Short-, intermediate-, and long-term tests of effects of ICS on
`HPA axis. Sensitivity for detecting systemic presence of ICS is greatest
`with AUC cortisol measurement; positive predictive value for risk of
`adrenal insufficiencv is increased bv dvnamic tests of adrenal resuon-
`,
`*
`
`siveness. UFC: urinary free cortisol.
`
`ones (BDP or BUD) when compared microgram-for-micro-
`gram. Such comparisons must, however, also consider
`clinically equivalent doses of ICS and account for delivery-
`device differences in lung drug deposition. Until the issue of
`equivalent dosage/device regimens is resolved, short-term
`microgram-for-microgram comparisons of ICS add little to the
`clinician’s ability to assess risk vs benefit of a particular drug.
`
`Failure to connect short-term results to
`long-term effect: bone metabolism
`
`Some studies of adults treated with ICS, for the most part in
`high doses, reveal reduced total bone mass and reduced bone
`mineral density (BMD) ( 5 ) , although excessive osteoporosis
`or fracture rates have not been reported in patients who have
`been treated with long-term ICS alone. An adverse effect of
`ICS on bone mineral accretion would be of particular
`concern in growing children, particularly during adoles-
`cence. The finding that levels of osteocalcin (61, serum
`markers of type 1 and type 111 collagen formation, and serum
`markers of type 1 collagen degradation (7, 8) can be affected
`by high-dose ICS treatment suggests that ICS have the
`potential to affect adversely bone growth, BMD, and
`eventual fracture risk, although the existing long-term
`evidence suggests that this is not the case (9, 10).
`Relationships between serum markers of bone metabolism
`and growth in ICS-treated patients are inconsistent. In children
`without asthma, the only prospective study of growth and
`serum osteocalcin showed no significant relationship ( 11 ).
`Studies of BDP- (12) and FP-treated ( 7 ) children found
`osteocalcin to be a poor marker for decreased growth (measured
`
`3
`
`

`

`Allen . Short-term studics of ICS
`
`Short-term
`
`,
`
`I
`
`Osteocalcin
`Type I procollagen
`Type 111 procollagen, Quantitative
`TvDe I C-terminal '
`*.
`CT
`,
`propeptide
`' turnover markers 1
`Urine bone
`Fra
`Positive predictive value for important adverse effect
`
`DXA
`
`QCT
`
`Figure 2. Short., intermediate-, and long-term tests of effects of
`ICS on bone. Markers of bone turnover have high sensitivity for
`detecting ICS systemic effect, but unverified positive predictive value
`for fracture risk or poor growth.
`
`term growth and pubertal development, and normal short-
`term (often abrupt) changes in growth rate. The difficulty of
`conducting controlled studies of ICS effects on long-term
`growth prompted interest in short-term (i.e., <6 months)
`studies using knemometry (measurement of
`lower leg
`growth) which could utilize: 1) very short observation
`periods; 2) controlled, randomized, double-blind conditions;
`3 ) crossover trials (18). Knemometry studies have generally
`shown growth suppression during treatment with BUD (by
`metered-dose inhaler) at doses of >400 pg/d, with BDP at
`doses of 2400 pg/d, and with BUD (Turbuhaler) or FP at 400
`%/day ( 1 9 1.
`Unfortunately, short-term lower leg-growth rates cannot
`be extrapolated to intermediate- or long-term growth.
`Childhood growth normally occurs in spurts, interspersed
`with periods during which essentially no growth occurs (20).
`The precision of knemometry detects this irregular pattern
`which, when combined with a focus on lower-leg growth
`only, actually becomes a liability in the prediction of overall
`linear body growth. Even when measurements are carried
`out over 4 months, 6-month growth velocity is predicted
`only within 28.8% (mean [z SO]), and correlations of growth
`of
`the lower leg with total height over 6 months are
`indifferent. Shorter time intervals produce even poorer
`predictions (21).
`The inability of knemometry to predict statural growth
`has been illustrated in one study of children treated with
`ICS ( 2 2 ) . In addition to normal variations in short-term
`growth, this result might also reflect differences in the
`consistency of ICS dose between short- and longer-term
`studies ( 2 3 ) . Recent well-designed and tightly monitored
`prospective studies reveal 12-month growth suppression
`
`during uninterrupted daily treatment with BDP 400 pg/
`day, a dose known to cause detectable growth slowing in
`knemometry studies (12, 24). Prospective studies which
`corn bine early application of knemometry with careful
`stadiometry measurements and close monitoring of
`consistent-dose IC administration over a period of 2 12
`months are needed to evaluate further the predictive value
`of knemometry measurements in IC-treated children.
`(Fig. 3 ) .
`The sensitivity of knemometry might allow the exclu-
`sion of risk for growth suppression by ICS based upon short-
`term screening for the absence of slowed lower-leg growth
`(19). However, reliable exclusion of an adverse effect based
`upon a negative test requires high specificity as well as
`sensitivity, and both aspects are compromised by normal
`short-term variations in growth. If a child begins ICS
`treatment during a time of normally slow or absent growth,
`knemometry measurements in subsequent weeks would
`not detect a slowing of growth attributable to ICS. Thus, for
`individual patients, knemometry studies have limitations
`in predicting both the presence and absence of long-term
`effects of ICS on growth. On the other hand, in the
`evaluation of study groups, knemometry appears to be
`useful in comparing differences in systemic activity
`between drugs and may be useful in identifying generally
`safe doses of ICS. Corroboration of short-term predictions
`of safe ICS doses with results of intermediate- and long-
`term studies is needed to strengthen the clinical value of
`knemometry studies.
`
`Sensitivity to detect systemic IC presence
`i Intermediate-term 1
`
`Short-term
`
`Lon
`
`Knemometry
`Bone markers
`Stadiometry
`(< 6 months)j
`
`Positive predictive value for important adverse effect
`-. - .
`--
`- -
`
`Figure 3. Short., intermediate-, and long-term tests of effects of ICS on
`growth. Highly sensitive knemometry is poor predictor of long-term
`growth, which is more accurately assessed by intermediate- or long-term
`stadiometry.
`
`4
`
`

`

`Conclusions: the added value of
`intermediate-term studies
`
`Several tests with high sensitivity for detecting systemic
`effects of ICS have been employed in short-term s t u l e s
`assessing the risks of IC treatment. The positive predictive
`value of
`these tests for the development of clinically
`important adverse effects is low. As illustrated above,
`AUC plasma cortisol profiles and urinary free-cortisol
`collections accurately indicate systemic presence of ab-
`sorbed ICS, but do not predict the loss of HPA axis
`rhythmicity. Markers of bone turnover can be correlated
`with changes in BMD (in adults) and growth rate, but have
`not been shown to predict either poor mineralization or
`growth in children treated with ICS. Knemometry sensi-
`tively assesses growth of the lower leg, but is confounded by
`normal week-to-week variations in childhood growth and
`correlates poorly with total body growth. Consequently,
`while short-term tests have value in detecting the presence
`and even the mechanism of a systemic effect of absorbed
`ICS, they do not, thus far, predict the consequences of these
`effects.
`Long-term s t u l e s of sufficient duration to evaluate actual
`occurrence of the potential clinically important adverse
`effects of ICS [episodes of adrenal insufficiency, reduced
`final adult height, increased fracture rate) are ultimately
`required to prove the risks of long-term ICS treatment.
`However, the time required for such studies inevitably
`introduces multiple confounding factors, particularly in a
`
`Allen . Short-term studies of ICS
`
`study group of patients with a chronic disease of fluctuating
`severity. For effects on linear growth and bone metabolism,
`the many years required to assess end points can even
`overlap periods of profound change in prevailing strategies
`for disease management.
`Although intermediate-term studies rarely provide infor-
`mation about the occurrence of clinically important adverse
`effects of ICS, they can evaluate physiologic changes
`directly connected [e.g., total body growth rate measured
`by careful stadiometry; diminished BMD by DXA) to the
`clinically important adverse effect. This offers an important
`advantage over short-term studies, in which reliance is
`placed on measurement of markers thought to correlate
`with the effect. Accordingly, the duration of adequate
`intermediate-term safety studies for ICS is defined by the
`time required to detect reliably this physiologic changej for
`example, estimated durations are of 2 3 months to demon-
`strate development of adrenal atrophy and diminished
`response to exogenous ACTH stimulation, of 2 1 2 months
`for the documentation of slowed growth in total height, and
`perhaps of 2 2 4 months for changes in BMD. Controlled
`studies of this duration are difficult to conduct, but recent
`efforts indicate that these obstacles can be overcome ( 2 s ) .
`Consequently, well-designed intermediate-term studies of
`the response of the HPA axis to stimuli, changes in BMD, or
`statural growth, which combine controlled study feasibility
`with improved positive predictive value, provide the best
`practical opportunity to increase knowledge about the long-
`term risks of ICS.
`
`References
`I . Expert panel report 2 . Guidelines for the
`diagnosis and management of asthma.
`Bethesda, MD: 1997, National Institutes of
`Health Publication No. 97-405 1.
`2. Allen DB. Inhaled corticosteroids in children.
`In: Middleton E, Reed CE, Ellis EF, Adkinson
`NF, Yunginger JW, Busse WW, editors.
`Allergy: principles and practice. 4th ed.
`St Louis, MO: Mosby Yearbook, 1993:Update
`15,6-16.
`3 . Nicolaizik WH, Marchant JL, Prcece MA,
`Warner 70. Endocrine and lung function in
`asthmatic children on inhaled
`corticosteroids. Am 1 Respir Crit Care Med
`1994;1s0:624-618.
`
`4. Lonnebo A, Grahnen A, Jansson B, Brundin
`RM, Ling-Andersson A, Eckemas SA. An
`assessment of the systemic effects of single
`and repeated doses of inhaled fluticasone
`propionate and inhaled budesonide in healthy
`volunteers. Eur I Clin Pharrnacol
`1996;49:459-463.
`5 . Toogood JH, Hodsman A, Baskerville J,
`Markov A. Vertebral bone mineral density
`among elderly men and women receiving
`long-term inhaled or oral steroids for asthma.
`f Allergy Clin lrnrnunol 1995;95:310.
`6. Soma R, Turpeinen M, funtunen-Backman K,
`Karonen SL, Sorva A. Effects of inhaled
`budesonide on serum markers of bone
`metabolism in children with asthma. [Allergy
`Clin Immunol 199~;90:808-8i5.
`
`7. Wolthers OD, Hansen M, Juul A, Nielsen HK,
`Pederson S. Knemometry, urine cortisol
`excretion, and measures of insulin-like
`growth factor axis and collagen turnover in
`children treated with inhaled glucocorticoids.
`Pediatr Res 1997;41:44-50.
`8. Crowley S, Trivedi P, Risteli L, Risteli J,
`Hindmarsh PC, Brook CGD. Collagen
`metabolism and growth in prepubertal
`children with asthma treated with inhaled
`steroids. I Pediatr i998;13~:409-413.
`9. Agertoft L, Pedersen S. Bone mineral density
`in children with asthma receiving long-term
`treatment with inhaled budesonide. Am
`Respir Crit Care Med i998;157:178-183.
`
`5
`
`

`

`Allen . Shurt-term studies of ICS
`
`10. Boot AM, de Jongste JC, Vcrberne AP, Pols
`HA, dc Muinck Keizer-Schrama SM. Bone
`mineral density and bone metabolism of
`prepubertal children with asthma after long-
`tcrm treatment with inhaled corticosteroids.
`Pediatr I’ulmonol 1 $197~24: 3 79-3 84.
`I 1. Michaclson KF, Johansen JS, Samuelson G,
`Price PA, Christiansen C. Serum bone
`gamma carboxyglutamic acid protein in a
`longitudinal study of infants; lower values in
`formula-fed infants. Pediatr Res i99zj3i:4io-
`4’5
`12. Doull IJM, Freczcr NJ, Holgatc ST. Growth of
`prepuhcrtal children with mild asthma
`treatcd with inhaled beclomethasone
`dipropionatc. A m 1 Respir Crit Care Med
`1995;1(1:1715-1719.
`I 3 . Konig P, Hillman L, Ccrvantcs C, et al. Bone
`metabolism in children with asthma treated
`with inhaled beclomethasone dipropionatc.
`[ Pediatr 1993;122:219-226.
`14. Doull I, Freezer N, Holgatc S. Ostcocalcin,
`growth, and inhaled corticostcroids: a
`prospective study. Arch Dis Child
`1996;74:497-50l
`
`1s. Agertoft L, Pedersen S. Final height of
`asthmatic children treated for 7-1 i years
`with inhaled budesonide. Am I Respir Crit
`Care Med 1998;156(4, pt 2):A71i.
`16. Schneider DL, Barrett-Connor EL. Urinary N-
`telopeptide levels discriminate normal,
`osteopenic, and osteoporotic bone mineral
`density. Arch Int Med 1997;i57:1~41-1245.
`17. Marshall D, Johnell 0, Wedel H. Meta-
`analysis of how well measures of bone
`mineral density predict occurrence of
`osteoporotic fractures. BMI 1996;)ix 1254-
`
`1259.
`18. Wolthers OD. Long-, intermediate-, and
`short-term growth studies in asthmatic
`children treated with inhaled
`glucocorticosteroids. Eur Respir 1996;9:821-
`827.
`19. Pedersen S, O’Byme P. A comparison of the
`efficacy and safety of inhaled corticosteroids
`
`in asthma. Allergy 1 9 9 7 ; ~ ~ (Suppl 39):1-34.
`20. Hermanussen M, Geitcr-Benoit K,
`Burmeister r, Sipplc WG. Periodical changes
`of short term growth velocity in human
`growth. Ann Hum Biol 1988;15:103-109.
`
`21. Walcs JKH, Milner RDG. Knemometry in
`assessment of linear growth. Arch Dis Child
`1987;62:166-171.
`2 2 . Agcrtoft L, Pcdcrscn S. Short-tcnn lower leg
`growth in children during treatment with
`fluticasonc and BUD: a dose response study.
`Am I Respir Crit Care Med
`1995;151(4, Pt 2):AlSo.
`2 3 . Lemanske RF, Allen DB. Choosing a long-
`term controller medication in childhood
`asthma: the proverbial two-edged sword. Am
`I Respir Crit Care Med 1997;156:685-687.
`24. Verberne AAPH, Frost C, Roorda RJ, van der
`Laag H, Kenibijn KF. One year treatmcnt
`with salmeterol compared to beclomethasonc
`in children with asthma. A m I Respir Crit
`Care Med 1997;156:688-695.
`2 5 . Allen DB, Bronsky EA, LaForcc CF, et al.
`Growth in asthmatic children treated with
`fluticasone propionatc. I Pediatr
`1998;132:472-477.
`
`6
`
`