Pediatric Pulmonology 43:945–952 (2008)
Original Articles
Annual Assessment Spirometry, Plethysmography,and Gas Transfer in Cystic Fibrosis:
Do They Predict Death or Transplantation
Mark Rosenthal, MD, FRCP, FRCPCH*
Summary. Aim: The long- and short-term prognostic value of pediatric spirometry, plethysmog-
raphy, and gas transfer measurements in cystic fibrosis (CF) were assessed. Methods: Two
hundred ninety-eight children with CFand�4 annual assessment lung function measurements at a
single institution were analyzed in mid childhood. Long-term outcome was death or lung
transplantation (D/T) before 2007. Short-term outcome was forced expired volume in one second
(FEV1) z-score 1 year after the previous lung function measurements. Results: 26/298 had a D/T
outcome at median 19.5 years. A zFEV1<�2 aged 8 years had a positive predictive value of 67%
(sensitivity 67%) for D/T in those homozygous for DF508 but zFEV1 at older ages and all geno-
types was unhelpful. The ratio of residual volume to total lung capacity z-score could also predict
a few D/T individuals when zFEV1 was normal in mid childhood. Most other lung function
measurements were not helpful. Matching D/T with alive groups for year of birth left prognostic
utility unchanged. Only current zFEV1 could significantly predict zFEV1 1 year hence (56%
variability explained, P<0.00001); no other lung function, gender, age or nutrition factor was
significant. Conclusion: The value of routine plethysmography and gas transfer measurements in
CF is questionable in CF management. Detecting abnormal spirometry even at age 8 years may be
too late to affect long-term outcome. Pediatr Pulmonol. 2008; 43:945–952. � 2008 Wiley-Liss, Inc.
Key words: cystic fibrosis; death; gas transfer; plethysmography; spirometry; trans-
plantation.
INTRODUCTION
It is axiomatic that, as part of an annual assessment(AA) in cystic fibrosis (CF), lung function measurementsare undertaken. Spirometry is always performed and inaddition the current (2001) UK CF trust standards of careguidelines state ‘‘. . . and when appropriate, more detailedlung function tests, depending on local availability,including lung volumes, CO transfer, exercise testingand challenge testing’’ (http://www.cftrust.org.uk/aboutcf/publications/consensusdoc/c_3000standards_of_care.pdf).There is little published evidence as to the value ofplethysmography and gas transfer in CF both for short-term evaluation and long-term prognosis. Liou’s model1
for predicting the chance of a patient surviving 5 yearsbased on 11,600 US patients had no plethysmographic orgas transfer measurements to include but did demonstratethat FEV1 was one of the less significant measures (andrate of change of FEV1 not significant) compared with thepresence of CF related diabetes as an example. Merkuset al.2 showed that in children at least, gas transfer was nota suitable early marker of disease progression. Nielson
et al.3 did demonstrate in a 4-year longitudinal pediatricstudy that specific total lung resistance (a plethysmo-graphic measurement) was more consistently abnormalthan for example total lung resistance measure by theinterrupter technique.
As plethysmography and gas transfer measurementsare time consuming, technically demanding and expen-sive, and the results usually not interpretable in routineclinical practice their long and short-term prognosticvalue was assessed in comparison with spirometry.
The Royal Brompton Hospital, Sydney Street, London, UK.
*Correspondence to: Mark Rosenthal, MD, FRCP, FRCPCH, The Royal
Brompton Hospital, Sydney Street, London SW3 6NP, UK.
E-mail: [email protected]
Received 5 March 2008; Revised 9 May 2008; Accepted 23 May 2008.
DOI 10.1002/ppul.20879
Published online 9 September 2008 in Wiley InterScience
(www.interscience.wiley.com).
� 2008 Wiley-Liss, Inc.
METHODS
The study was carried out in one tertiary CF center, TheRoyal Brompton Hospital, London, UK. The pediatricclinic has kept a patient database which includes AA lungfunction measurements since 1985. Currently 731 childrenare registered of whom 394 were born prior to 01.01.1993.298/394 had had at least 4 AA lung function measure-ments at any time in this clinic and were on the databaseand it these 298 that form the study group. There were noexclusion criteria. Only lung function measurements‘‘formally’’ measured in our lung function laboratoryrather than routine clinic spirometry were used. Spirome-try measured forced vital capacity (FVC), forced expiredvolume in one second (FEV1), peak expiratory flow (PEF)and measures of flow at lower lung volumes. Plethysmog-raphy measured total lung capacity (TLC), residual volume(RV) and where technically possible specific lung resis-tance (Sraw), gas transfer measurements measured carbonmonoxide transfer (Tlco) and accessible lung volume(VA). Inevitably where patients transferred in or out fromother centers or failed to attend or perform technicallyreliable results, incomplete data would be present.
Z-scores ((actual result�predicted result)/populationstandard deviation) based on normal data4,5 were used toeliminate the inevitable effects of age, height, and genderon lung function. By definition a normal population has amean z-score of zero and 95% of values lie between þ2and �2. The original normal data5 did not present RV/TLC and was thus reanalyzed so that zRV/TLC could becalculated in the present study. A basic measure ofnutrition used the body mass index z-score (zBMI).
The long-term outcome measure was death or trans-plantation (D/T) by January 4th 2007. The short-termoutcome measure was zFEV1 1 year after the previous AAlung function. Measurements from 8 to 12 years of agewere particularly examined for several reasons: theywere technically reliable; patients were young enough toenable intervention should a lung function measurementbe a realistic indicator of early death/transplantation(see Discussion Section); it was at least 1 year beforethe earliest death in this cohort and thus avoided datacensoring.
To maintain a strict longitudinal analysis and maximizeits practical usefulness (cf mixed effect modeling), astraightforward measure of positive predictive value(PPV) and sensitivity (sens) was determined for allmeasures alone or in combination for children aged 8–12 years. To determine the best balance between PPVand sensitivity, they were plotted against each other atvarious lung function z-score cut-offs similar to a receiveroperating curve (ROC) but with the advantage of adjustingfor the background rate of occurrence.6 There are no setcriteria for what a useful PPVor sensitivity should be as itdepends on the importance of the outcome and theconsequences of missing a case or incorrectly identifyinga case. For a crucial outcome such as (D/T), a highsensitivity is essential so no cases are missed but the PPValso has to be high otherwise interventions may occurwhich are not warranted as the test cannot discriminate.Thus, a PPV of 70% and a sensitivity of >90% werearbitrarily chosen as minimum requirements. For compar-ison, tossing a coin for an outcome likelihood of 10%,gives a PPVof 5% and a sensitivity of 50%.
For the short-term outcome of zFEV1 1 year hence,multiple stepwise linear regression used the factors:current zFEV1, zVA, zTlco, zKco, zBMI, zRV/TLC, ageat test, genotype (DF508 homozygous vs. the rest),and year of birth, this last being a proxy for treatmentintensity and sophistication. As children had up to fivemeasurements between 8 and 12 years only their first twomeasurements a year apart were used to avoid the effectsof repeated measures. This occurred in 276/298 children.
RESULTS
26/298 (8.7%) were (D/T) by median age 19.4 years.From Table 1, D/T subjects were more likely to behomozygous DF 508 (P¼ 0.014) and born some 5 yearsearlier than those still alive (P< 0.001).
Long-Term Outcome—Death/Transplantation
Tables 2a and 2b show the allocation of subjects tocategories (alive or D/T) and z-scores<�2 or not (>þ2 ornot for RV/TLC) alone and in combination together with
TABLE 1— Cohort Characteristics (Median and Range Unless Stated)
Total Alive Dead/transplant Significance
Number 298 272 26
Gender 151M/147F 138M/134F 13M/13F 0.94
Year of birth 1987 (1977–1993) 1987 (1977–1993) 1982 (1978–1992) <0.0001
DF508/DF508 160 (54%) 140 (51%) 20 (77%) 0.014
DF508/known 43 41 2
DF508/unknown 46 47 1
Known/known 9 8 1
Known/unknown 4 4 0
Unknown/unknown 36 34 2
Age at death/trans 19.4 (13.2–26)
Pediatric Pulmonology
946 Rosenthal
TA
BL
E2a—
(All
Su
bje
cts
):P
osit
ive
Pre
dic
tive
Valu
es
(PP
V%
)an
dS
en
sit
ivit
ies
(Sen
s%
)fo
rzF
EV
1,zR
V/T
LC
,zV
A,zB
MIan
da
Co
mb
inati
on
Sco
re(S
co
res
1if
Eit
her
zF
EV
1<�
2,
or
zR
V/T
LC>
2,
or
zV
A<�
2an
dth
eR
esu
ltS
um
med
)
Cu
t-o
ff
8y
ears
9y
ears
10
yea
rs1
1y
ears
12
yea
rs
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
zFE
V1
<�
28
74
7/7
02
64
13
/33
40
13
25
/68
40
13
25
/76
77
91
0/6
4
��
24
33
14
88
15
56
15
94
14
15
zRV
/TL
C>þ
29
53
6/5
03
76
14
/50
45
14
24
/82
51
14
15
/82
77
91
0/6
4
�2
42
51
32
61
44
31
44
33
75
ZF
EV
1<�
2o
r
zRV
/TL
C>
2
<�
2o
r
>2
11
73
9/7
04
17
15
/58
55
17
24
/89
60
15
14
/88
10
21
09
/71
Th
ere
st4
03
12
95
13
72
13
72
11
64
ZB
MI
<�
22
13
3/9
40
0/0
10
00
/01
21
7/6
12
32
0/2
0
��
24
91
01
72
14
18
71
91
88
16
20
91
2
zVA
<�
26
22
5/2
03
26
7/1
81
06
38
/33
13
73
5/7
01
52
12
/14
��
24
58
16
99
18
31
21
83
10
19
61
2
Co
mb
inat
ion
sco
re(s
ee
leg
end
)
03
93
13
15
13
82
13
42
11
54
15
32
36
28
63
23
47
3
23
34
2/4
52
11
12
/21
25
52
6/5
82
85
26
/71
50
61
41
2/5
3
34
22
27
66
79
21
8
Max
imu
msc
ore¼
3,r
ang
e0
–3
.Ex
amp
le:ag
ed1
0y
ears
,rea
din
gh
ori
zon
tall
y5
5/7
2su
bje
cts
had
aZ
FE
V1<�
2o
rzR
V/T
LC>
2an
dw
ere
aliv
ean
dth
eo
ther
17
wer
ed
ead
/tra
nsp
lan
ted
(PP
V1
7/
72¼
24
%).
Rea
din
gver
tica
lly,
of
the
19
(17þ
2)
D/T
sub
ject
ste
sted
aged
10
yea
rs,1
7/1
9h
ada
ZF
EV
1<�
2o
rzR
V/T
LC>
2an
dth
eo
ther
2d
idn
ot(s
ensi
tiv
ity¼
17
/19¼
89
%).
Ag
ed1
0y
ears
,5
had
aco
mb
inat
ion
sco
reo
f2
and
6a
sco
reo
f3
.T
he
PP
V/s
ensi
tiv
ity
of
asc
ore
of
2o
rm
ore¼
26
%an
d5
8%
resp
ecti
vel
y.
TA
BL
E2b
—D
F508
Ho
mo
zyg
ou
sO
nly
Cu
t-o
ff
8y
ears
9y
ears
10
yea
rs1
1y
ears
12
yea
rs
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
Ali
ve
D/T
PP
V/s
ens
zFE
V1
<�
23
66
7/6
71
62
11
/22
21
72
5/5
42
11
13
4/7
94
19
18
/69
��
23
13
83
78
76
87
37
54
zRV
/TL
C>þ
24
45
0/3
32
45
17
/50
25
10
29
/77
26
12
32
/86
44
92
0/6
9
�2
30
87
45
81
38
12
76
4
ZF
EV
1<�
2o
r
zRV
/TL
C>
2
<�
2o
r
>2
56
55
/67
25
51
7/5
03
01
22
4/8
63
11
32
3/7
15
81
01
5/7
7
Th
ere
st2
93
73
57
82
77
16
43
zBM
I<�
20
00
/02
00
/08
00
/08
11
1/7
11
32
1/6
3
��
23
49
10
01
21
01
14
10
21
31
19
5
zVA
<�
26
22
5/2
03
26
7/2
21
06
38
/33
13
73
5/4
11
52
12
/14
��
24
58
16
99
18
31
21
83
10
19
61
2
Co
mb
inat
ion
sco
re(s
ee
bel
ow
)
02
83
74
57
92
76
16
33
13
21
35
12
51
63
26
1
22
35
7/4
41
31
12
/17
12
42
5/4
61
55
36
/71
31
62
0/6
7
31
12
16
23
54
2
Res
tas
for
Tab
le2
a.
Pediatric Pulmonology
Lung Function and Prognosis in Cystic Fibrosis 947
PPV and sensitivity (sens) for each year from 8 to 12 forall subjects (top panel) and only those homozygous forDF508 in the bottom panel. PPVs varied from 0 to amaximum of 47% and sensitivities from 14% to 89%. Nomeasurement alone or in combination at any age had aPPVand sensitivity similar to the desired 70% PPV/>90%sensitivity. The most consistently useful test across theage range studied was zFEV1. There was tendency for allmeasures to be less useful at ages 11 and 12 yearscompared with 8–10 years. Figure 1 shows as anexample, the relationship in 10-year olds of PPV against1-sensitivity for FEV1 and VA for z-score cut-offs of �2,�2.5, and �3 and for RV/TLC z-score cut-offs of þ2,þ2.5, and þ3.
Because of the differing years of birth between thealive and the D/T groups, data for homozygous DF508children age 10 years were reanalyzed using only subjectsborn before June 30th 1987. As an example, the PPVand sensitivities for death/transplantation respectivelyat 10 years for zFEV1<�2 were 43% and 58%, zRV/TLC> 2 were 50% and 72%, and for zVA<�2, PPVwere 33% and sensitivity 17%.
To determine if zRV/TLC provided additional utility tojust measuring zFEV1 alone, Figure 2 shows a panel ofscatter plots for zFEV1 against zRV/TLC for each yearfrom 8 to 12 years. Reference lines are drawn at �2 forzFEV1 and þ2 for zRV/TLC. The top right quadrant ofeach panel therefore represents the situation of a zFEV1 in
the normal range but a raised zRV/TLC. At age 10 yearsfor example, 19 subjects had a normal zFEV1��2 but araised zRV/TLC> 2. 4/19 were destined to D/T a median9 years later and would not have been detected on the basisof zFEV1 alone aged 10 years. 18/19 were retested aged11 years. 11/18 still had a zRV/TLC> 2 of whom 2/11would D/T a median 9 years later but now 5/11 also had azFEV1<�2 of whom 1/5 would die. The 4th D/T subjectwas not tested at 11 years. The remaining 7/18 now hada normal zRV/TLC, a normal FEV1 and 1/7 would die amedian 9 years later.
Using a methodology that would only measure RV/TLCif FEV1 was normal, 158 such tests would be performedto find 19 subjects with a raised RV/TLC, 4 of whom(ppv 22%, sensitivity 67%) would be destined to D/T amedian 9 years later. 15/19 would be wrongly categorizedand 2/14 subjects destined to D/T would remain missed atage 10 years as they had a normal RV/TLC and FEV1.
To determine an intermediate outcome, 36/138 subjectstested aged 15 years had a zFEV1��2.00 and a zRV/TLC> 2 at either 10, 11, or 12 years and 102/138 azFEV1��2.00 and a zRV/TLC� 2, that is, were normal.At 15 years, there was a trend but no significant differencein zFEV1 between these two groups (mean zFEV1 �1.7vs. �1.3, P¼ 0.08, Mann–Whitney).
ZVA detected no subjects aged 10 years destined to diewith a concurrent normal zFEV1.
Table 3 shows rate of change data over successive2-year periods for lung function parameters and zBMI forall subjects (top panel) and only those homozygous forDF 508 (bottom panel). Rates of change for zFEV1, zRV/TLC, zVA, and zBMI were very similar between the aliveand D/T groups. There was a significantly greater fallin zKco in the D/T group between 9 and 11 years and also10 and 12 years. The PPV for a zKco fall of>0.5 was 18%and sensitivity 67%. Few if any children aged 8–12 yearsin the D/T group had a zBMI<�2 or had a steeper declinethan their alive counterparts.
Short-Term Outcome—zFEV1 1 Year Hence
Multiple linear regression on 276 subjects demon-strated that only zFEV1 at baseline was significantlyrelated to outcome in children aged 8–12 years inclusive,and even though highly significant (P< 0.0001) onlyexplained 53% of the variability of the outcome (56% ifhomozygous DF508 only). No other factor was signifi-cant. Using all the other variables only added a further0.2% to the variance explained. Using the entire age range(5–19 years) the percent variability explained was 66%with the other factors contributing a further 1%.
DISCUSSION
The CF AA serves many functions, principally anopportunity to review the patient from every standpoint
Fig. 1. The relationship of positive predictive value and
1-sensitivity for death/transplantation for zFEV1 zRV/TLC and
zVA for children aged 10 years relating positive predictive value
and sensitivity at various z-score cut-offs.
Pediatric Pulmonology
948 Rosenthal
and organ system and thus maximize treatment effective-ness and to provide the ‘‘space’’ for discussion of longer-term issues such as prognosis, fertility, careers, treatmentadherence or any other subject occupying the patient orparent’s mind. The time commitment from the familyis considerable as in some centers, the assessment is awhole day event followed by a review of results afew weeks later. Personal experience shows that thecommonly verbalized question at AA of ‘‘how are theydoing compared with the average’’ has a less commonlyverbalized sub-text ‘‘how long are they going to live.’’To answer the question accurately and for the AA toretain credibility, the tests undertaken must have eithershort- or long-term utility or both.
From this study, if a subject has a zFEV1 in the normalrange, the chance of the zRV/TLC being abnormal isbetween 11% and 18% depending on age (8–12 years). Isknowing this fact helpful? The current study does notprovide evidence that it is, given that most of those in thatgroup destined to D/Tyoung would be picked up by zFEV1
by the next year and that lung function at 15 years was notsignificantly different depending on whether the zRV/TLC was normal or not. A possible alternative conclusionis that an abnormal RV/TLC may be an early warning sign
of a poor outcome in a small proportion of individuals witha normal FEV1 but its PPV (22%) is weak.
Does knowing the zRV/TLC or zFEV1 enable one toanswer the family’s question ‘‘how long will he live’’?Lung function between the 8 and 12 years was a poorindicator of the risk of dying or being transplanted youngeven in those homozygous for DF508. This confirms thework of others in different contexts. Kerem’s seminal 1992paper7 relating an FEV1 <30% to a 50% chance of deathwithin 2 years was effectively only as good as tossinga coin. Milla’s study8 of only subjects with an FEV1 of<30% showed an even poorer association with death.An Australian study9 showed that median FEV1 2 yearsbefore a pediatric CF death was 53% but the range was24–102% with a significant gender effect, boys having amedian of 69% compared with females’ 47%.
It may not be entirely surprising that the ability topredict an event nearly 10 years hence would be sopoor given the number of confounders involved such asacquisition of new micro-organisms, treatment adherenceduring adolescence, the effects of newly introducedmedications, the increasing aggressiveness of treatmentand the lowering of thresholds for such treatment. A proxyfor the last three confounders may be year of birth as
Fig. 2. The relationship between zFEV1 and zRV/TLC at each year from age 8–12 years. Gridlines
set at the limits of normality (�2 for zFEV1 and þ2 for zRV/TLC). þ¼Patients destined to die or
have lung transplants early (D/T); *¼ those still alive.
Pediatric Pulmonology
Lung Function and Prognosis in Cystic Fibrosis 949
TA
BL
E3—
Ch
an
ges
inz-S
co
reV
alu
es
(Mean
,95%
CI)
Over
the
Pre
vio
us
2Y
ears
8yea
rs9
yea
rs10
yea
rs11
yea
rs12
yea
rs
Ali
ve
D/T
Ali
ve
D/T
Ali
ve
D/T
Ali
ve
D/T
Ali
ve
D/T
All
gen
oty
pes
No
51
11
177
14
198
19
200
17
221
15
Gen
der
M/F
25/2
63/8
86/9
15/9
100/9
812/7
101/9
98/9
109/1
12
9/6
Yea
rof
bir
thm
edia
n1985
1984
1989
1985***
1989
1984****
1989
1984****
1989
1985***
Chan
ge
inzF
EV
1over
pre
vio
us
2yea
rs
�0.1
5
(�0.6
4to
0.3
4)
�0.1
9
(�1.3
5to
�0.9
8)
�0.2
6
(�0.4
2to
�0.1
0)
�0.4
7
(�2.0
1to
�1
.08
)
�0.3
1
(�0.4
8to
�0.1
4)
�0.6
6
(�1.2
9to
�0.0
2)
Chan
ge
inzR
V/T
LC
over
pre
vio
us
2yea
rs
�0.0
16
(�0.0
43
to0.0
11)
0.0
15
(�0.0
84
to0.1
12)
�0.0
4
(�0.1
1to
0.0
2)
0.0
5
(�0.0
3to
0.1
3)
�0.0
09
(�0.0
19
to0.0
02)
�0.0
11
(�0.0
53
to0.0
32)
Chan
ge
inzB
MI
over
pre
vio
us
2yea
rs
�0.1
8
(�0.5
1to
0.1
6)
�0.1
3
(�1.4
7to
1.2
1)
�0.1
8
(�0.3
9to
0.0
4)
�0.7
1
(�1.5
9to
0.1
8)
�0.0
2
(�0.2
3to
0.1
9)
0.3
0
(�0.5
7to
1.1
6)
Chan
ge
inzK
coover
pre
vio
us
2yea
rs
0.0
1
(�0.3
4to
0.3
6)
0.6
2
(�0.6
5to
1.8
8)
0.0
6
(�0.1
1to
0.2
4)
�0.6
6{
(�1.4
9to
0.1
7)
0.1
2
(�0.0
1to
0.2
6)
�1.1
5****
(�1.7
0to
�0.6
1)
Chan
ge
inzV
Aover
pre
vio
us
2yea
rs
0.2
8
(�0.2
4to
0.8
1)
�0.2
0
(�0.9
6to
0.5
5)
0.0
8
(�0.1
2to
0.2
9)
�0.2
5{
(�1.4
6to
0.9
7)
�0.0
3
(�0.2
2to
0.1
6)
0.4
1
(�0.4
0to
1.2
)
DF
508/D
F508
No
34
9102
12
109
14
110
14
124
14
Chan
ge
inzF
EV
1over
pre
vio
us
2yea
rs
�0.6
0
(�1.0
0to
�0.2
0)
�0.0
4
(�1.1
9to
�1.2
8)
�0.2
1
(�0.4
0to
�0.0
3)
�0.8
4
(�2.7
0to
�1
.04
)
�0.3
4
(�0.5
6to
�0.1
1)
�0.7
9
(�1.4
1to
�0.1
6)
Chan
ge
inzR
V/T
LC
over
pre
vio
us
2yea
rs
�0.0
0
(�0.0
3to
0.0
2)
0.0
0
(�0.1
1to
0.1
0)
�.0
8
(�0.1
9to
0.0
3)
0.0
4
(�0.0
7to
0.1
5)
0.0
00
(�0.0
12
to0.0
12)
�0.0
12
(�0.0
58
to0.0
35)
Chan
ge
inzB
MI
over
pre
vio
us
2yea
rs
�0.2
1
(�0.6
1to
0.1
7)
�0.0
2
(�1.5
9to
1.5
5)
�0.3
2
(�0.6
1to
�0.0
3)
�1.0
5
(�2.0
2to
�0
.07
)
0.2
3
(�0.0
3to
0.4
9)
0.5
2
(�0.2
4to
1.3
0)
Chan
ge
inzK
coover
pre
vio
us
2yea
rs
0.2
1
(�0.2
1to
0.6
3)
0.6
3
(�0.8
9to
2.1
4)
�0.0
8
(�0.3
0to
0.1
4)
�0.6
0
(�1.9
0to
0.7
1)
0.0
8
(�0.1
1to
0.2
7)
�1.0
6****
(�1.6
2to
�0.5
0)
Chan
ge
inzV
Aover
pre
vio
us
2yea
rs
0.0
3
(�0.5
5to
0.6
1)
�0.1
1
(�0.9
8to
�0.2
1)
0.0
7
(�0.2
1to
0.3
6)
�0.4
3
(�2.0
6to
1.1
9)
0.0
1
(�0.2
5to
0.2
7)
0.1
8
(�0.5
2to
0.8
9)
Th
e1
0-y
ear
colu
mn
show
sth
ech
ang
esi
nce
thei
rre
sult
aged
8y
ears
,an
dth
e1
1-y
ear
colu
mn
show
sth
ech
ang
esi
nce
thei
rre
sult
aged
9y
ears
,et
c.
**
*P¼
0.0
04
–0
.00
1b
yM
ann
–W
hit
ney
U-t
est.
**
**
P¼<
0.0
01
by
Man
n–
Wh
itn
eyU
-tes
t.{ P
<0
.01
–0
.05
by
Man
n–
Wh
itn
eyU
-tes
t.
P¼
0.0
06
by
Man
n–
Wh
itn
eyU
-tes
t.
Pediatric Pulmonology
950 Rosenthal
more interventions would be available to those bornmore recently. As an example, of 44 children tested aged12 years who at age 10 years had a zFEV1<�2, 11/44 at12 years had a zFEV1 better than �2 but 33/44 still had azFEV1<�2. The median date of birth was identical at1987.5 for each group implying that greater availabilityof interventions had no apparent bearing on the chance ofan improved zFEV1.
However the key difference between those alive and theD/T group was their year of birth with the latter group bornsome 4–5 years earlier. Given the median year of D/Twas19.5 years and the median age as of January 2007 ofthose alive and not transplanted was 16.5 years, this biasmay be the explanation for the poor prognostic quality oflung function, that is, those adults still alive who had poorlung function aged 8–12 years have ‘‘not had enough timeto die yet.’’ However in the D/T group, only three wereborn since the beginning of 1985, thus the probability ofthe still alive group suddenly changing category in the next3 years is very unlikely and if the predicted number ofpatients did do so, the PPV for zFEV1<�2 would onlyrise from 25% to 28%. Matching the cohorts for date ofbirth had very little effect either.
Just as lung function had a poor long-term prognosticutility, its short-term utility was also poor and that no otherfactor apart from current zFEV1 predicted zFEV1 1 yearhence. The argument that its utility was poor, becauseinterventions occurred to correct it, is not borne out by theexample cited above. One factor deliberately not includedin the risk factors was microbiological status because ofthe unreliability of the data and the problems of definingeither the date of acquisition or ‘‘chronic colonization’’ offor example Pseudomonas aeruginosa. It is accepted thataccurate knowledge of this may improve lung functionutility. In our center we have never had more than 2% ofsubjects with either Burkholderia Cepacia or MethicillinResistant Staphylococcus Aureus (MRSA).
The obvious drawbacks to this study is that despitebeing longitudinal, it is retrospective and drawn fromonly one center. It is not possible to be certain whetherthe findings would translate to other centers althoughthe literature cited above does not demonstrate a goodrelationship between prognosis and lung function. Never-theless one advantage of a single center study is theconsistency of measurements and practice. The othermajor consequence of such a study is that interventionswill have occurred as a consequence of the results thusdistorting the findings. This is undoubtedly true forspirometry as a poor result always mandates a response.However somewhat embarrassingly in our institution,the plethysmography and gas transfer results have beenroutinely ignored and RV/TLC will hardly ever haveinitiated an intervention. Indeed one purpose of this studywas to determine whether having ignored the results for solong we should continue performing them.
The final problem is that of timeliness. By choosing anindisputable long-term outcome (D/T), the time needed toelapse for this to occur renders the measurement periodso far in the past (10 yearsþ) that in the arena of rapidlyevolving CF care, the applicability of the findings becomequestionable. However using proxy outcome measures,for example lung function aged 18 years, is equallyproblematic as it assumes it really is closely related to thetrue outcome (D/T) which is the function of this study inthe first place—the relationship of lung function toprognosis.
Despite the above caveats, the study’s findings couldbe used to offer two differing conclusions. One is thatthe additional information provided by gas transfer andplethysmography does not support its continued routineannual measurement in children with CF because theirlong and short-term utilities are so poor. Sadly evenspirometry as judged by FEV1 did not assist long-termprognosis either but it should not also be discarded as itremains a useful tool for defining short-term changes inclinical status triggering diagnostic endeavor and clinicalintervention. However even FEV1 was hardly better at pre-dicting itself in 1 year time compared with tossing a coin.
The alternative conclusion is that RV/TLC shouldbe measured in those with a normal FEV1 as it providessome additional information given the importance of theoutcome. There is little point in measuring it if the FEV1
is abnormal as intervention is bound to occur. Despitethe significant rate of change of zKco, its ability todiscriminate was insufficient to warrant routine gastransfer measurements.
The most worrying possibility is that by the time onereaches mid childhood (8–12 years) when these measure-ments become technically reliable, those destined to‘‘do badly’’ are so categorized before measurements areusually possible, namely before aged 5 years. This issuggested by the declining overall ability of lung functionto predict D/T with increasing age. This is also borne outby US data10 showing that only lung function at 5 yearswas the prime difference between those dying young andthe rest. European data looking at long-term lung functionin CF subjects with and without allergic bronchopulmo-nary aspergillosis (ABPA) serendipitously showed thatrates of lung function decline in those clinically judged ashaving ‘‘mild, moderate or severe’’ CF disease were verysimilar irrespective of ABPA status but that their originallung functions at entry to the database were progressivelylower with increasing severity.11 The current data showan even poorer predictability at age 12 years comparedwith 8 years (Table 2a/b).
This behoves the CF community to validate a lungfunction measurement feasible in infancy for which thelung clearance index (LCI) is very promising. The LCI canbe undertaken with patience in any age group12 and has aconstant upper limit of normal irrespective of age and
Pediatric Pulmonology
Lung Function and Prognosis in Cystic Fibrosis 951
gender up to and including adulthood. It has been clearlyshown to be abnormal in a proportion of even asympto-matic CF infants and that intervention again does notnecessarily normalize it.13 Of course it remains unknownwhether LCI has any long-term utility either, howeverrather than add this to plethysmography and gas transfermeasurements at AA, it should replace them as part of along-term study.
REFERENCES
1. Liou TG, Adler FR, Fitzsimmons SC, Cahill BC, Hibbs JR,
Marshall BC. Predictive 5 year survivorship model of cystic
fibrosis. Am J Epidemiol 2001;153:345–352.
2. Merkus PJ, Govaere ES, Hop WH, Stam H, Tiddens HA, de
Jongste JC. Preserved diffusion capacity in children with cystic
fibrosis. Pediatr Pulmonol 2004;37:56–60.
3. Nielson KG, Pressler T, Klug B, Koch C, Bisgaard H. Serial lung
function and responsiveness in cystic fibrosis during early
childhood. Am J Respir Crit Care Med 2004;169:1209–1216.
4. Rosenthal M, Bain SH, Cramer D, Helms P, Denison D, Bush A,
Warner JO. Lung function in caucasian children aged 4–19 years.
I-Spirometry. Thorax 1993;48:794–802.
5. Rosenthal M, Cramer D, Bain SH, Denison D, Bush A, Warner
JO. Lung function in white caucasian children aged 4–19 years.
II-Single breath analysis and plethysmography. Thorax 1993;48:
803–808.
6. Altman DG. Practical statistics for medical research. London:
Chapman and Hall; 1991.
7. Kerem E, Reisman J, Corey M, Canny GJ, Levison H. Prediction
of mortality in patients with cystic fibrosis. New Engl J Med
1992;326:1187–1191.
8. Milla CE, Warwick WJ. Risk of death in cystic fibrosis patients
with severely compromised lung function. Chest 1998;113:1230–
1234.
9. Robinson PJ, Meehan J, Sawyer SM, Vyas J. Forced expiratory
volume in 1 second is an unreliable marker of risk of death in the
next 2 years. Pediatr Pulmonol 1999;19:A452.
10. Schechter MS, Shelton BJ, Margolis PA, Fitzimmons SC. The
association of socio-economic status with outcomes in cystic
fibrosis patients in the United States. Am J Respir Crit Care Med
2001;163:1331–1337.
11. Mastella G, Rainisio M, Harms HK, Hodson ME, Koch C,
Navarro J, Strandvik B, McKenzie SG. Allergic bronchopulmo-
nary aspergillosis in cystic fibrosis. A European epidemiological
study. Eur Respir J 2000;16:464–471.
12. Ranganathan SC, Dezateux C, Bush A, Carr SB, Castle RA, Madge
S, Price J, Stroobant J, Wade A, Wallis C, Stocks J, London
Collaborative Cystic Fibrosis Group. Airway function in infants
newly diagnosed with cystic fibrosis. Lancet 2001;358:1964–
1965.
13. Ranganathan SC, Stocks J, Dezateux C, Bush A, Wade A, Carr S,
Castle R, Dinwiddie R, Hoo AF, Lum S, Price J, Stroobant J,
Wallis C. The evolution of airway function in early childhood
following clinical diagnosis of cystic fibrosis. Am J Respir Crit
Care Med 2004;169:928–933.
Pediatric Pulmonology
952 Rosenthal
Top Related