Download Correlation of Serial Scleral and Corneal Pneumatonometry

Correlation of Serial Scleral and
Corneal Pneumatonometry
Debbie S. Kuo, MD,1 Yvonne Ou, MD,1 Bennie H. Jeng, MD, MS,2 Robert Bhisitkul, MD, PhD,1
Jay M. Stewart, MD,1 Jacque L. Duncan, MD,1 Ying Han, MD, PhD1
Purpose: To evaluate the usefulness of scleral pneumatonometry as an alternative for corneal measurements
of intraocular pressure (IOP) over a broad range of IOPs.
Design: Prospective, observational cohort study.
Participants: The study was conducted in the University of California, San Francisco, Retina Clinic between
August and November 2013 in 33 adult patients (age range, 34e94 years; mean standard deviation, 74.113.4
years) receiving antievascular endothelial growth factor intravitreal injections, which transiently increase IOP.
Methods: Corneal pachymetry and serial corneal and temporal scleral pneumatonometry (baseline, immediately after, and 10, 20, and 30 minutes after injection) were collected. One-time baseline corneal and scleral
pneumatonometry readings were obtained in the noninjected eye.
Main Outcome Measures: Correlation analysis and a Bland-Altman plot were used to evaluate reliability and
agreement between scleral and corneal measurements of IOP. A linear mixed model was used to determine the
relationship between measurements and to perform covariate analyses.
Results: Scleral and corneal pneumatonometry showed nearly 1:1 linear correlation, although scleral
pneumatonometry was biased toward higher values (r ¼ 0.94; P < 0.001). Scleral pneumatonometry averaged 9.0
mmHg higher than corneal pneumatonometry (95% limits of agreement, 1.5 to 19.5 mmHg). A linear mixed
model resulted in the following equation: corneal IOP ¼ 1.04 scleral IOP 10.37. Age, central corneal thickness, laterality, and glaucoma and lens status did not impact this relationship. The difference between corneal
and scleral pneumotonometry was correlated between the two eyes of individual patients (r ¼ 0.75; P < 0.001).
Conclusions: Differences between serial scleral measurements reflect differences between serial corneal
measurements. Scleral pneumatonometry should be considered as an alternative to corneal pneumatonometry for
following patients in whom corneal measurements are unreliable or unobtainable. Ophthalmology 2015;122:17711776 ª 2015 by the American Academy of Ophthalmology.
Intraocular pressure (IOP) normally is measured over the
cornea. However, for patients with significant corneal
pathology, such as scarring, thinning, and edema, or for
those who have keratoprosthesis implants, corneal tonometry can be inaccurate or impossible to obtain. However,
these corneal diseases are associated commonly with either
primary or secondary glaucoma. For example, in the case
of keratoprosthesis, difficulty with IOP measurement is a
significant problem. Glaucoma has been reported to be a
preoperative comorbidity in more than two-thirds of patients
and to be newly diagnosed in an additional 13% to 25%
of patients after keratoprosthesis implantation.1e3 Furthermore, keratoprostheses are associated with postoperative
elevation in IOP and progression of glaucoma, which can
become vision limiting.1e3
Scleral pneumatonometry has been proposed as an
alternative method for IOP measurement in patients for
whom corneal measurements are not possible. In a study
performed in cadaveric eyes, we previously showed that
serial measurements of scleral pneumotonometry correlate
strongly and linearly to IOP when IOP was set from 20 to 50
mmHg by infusion cannula.4 Importantly, this relationship
was unchanged after the eyes underwent keratoprosthesis
2015 by the American Academy of Ophthalmology
Published by Elsevier Inc.
implantation. In patients, a cross-sectional study by
Kapamajian et al5 found a positive correlation between
one-time corneal and scleral pneumatonometry in healthy
adult patients. However, the IOP range was limited by the
physiologic pressures of this population (10.5e27 mmHg),
and the relationship between changes in corneal and
scleral pneumatonometry in patients was not studied.
Furthermore, scleral pneumatonometry generally resulted in
higher measurements than corneal pneumatonometry, but
this difference was highly variable across individuals (mean
standard deviation, 8.45.7 mmHg).5
For scleral pneumotonometry to be a useful clinical tool,
scleral measurements should correlate to corneal measurements over a wide range of both physiologic and pathological pressures and have a predictable relationship over
multiple measurements when used to follow patients clinically. Therefore, in the current study, we measured serial
scleral and corneal pneumatonometry in patients receiving
intravitreal injections, which transiently increase IOP, to
evaluate the relationship between these 2 measurements
over a broad range of IOPs. Since the baseline difference
between scleral and corneal pneumatonometry in an eye
with corneal disease may be unknown, in the case of
http://dx.doi.org/10.1016/j.ophtha.2015.05.033
ISSN 0161-6420/15
1771
Ophthalmology Volume 122, Number 9, September 2015
unilateral or asymmetric disease, we hypothesized that one
could use the contralateral eye as a surrogate for the baseline
difference in the eye of interest. Thus, we also evaluated
whether the difference between corneal and scleral measurements was correlated between the 2 eyes of individual
patients.
Methods
Study Design
The Institutional Review Board/Ethics Committee at University of
California, San Francisco, approved this prospective observational
study. This study complied with Health Insurance Portability and
Accountability Act regulations and adhered to the tenets of the
Declaration of Helsinki.
Adult patients receiving antievascular endothelial growth factor intravitreal injections in the Retina Clinic of the University of
California, San Francisco, were recruited between August and
November 2013. We had a minimum target enrollment of 28 patients, which was predicted to have a 90% power to detect a correlation coefficient of 0.57 (based on the results from Kapamajian
et al5) with an a of 0.05 in an a priori sample size calculation.
Patients with previous incisional glaucoma surgery, scleral
buckle, strabismus surgery, refractive cornea surgery, scleral
pathology such as thinning or scarring, or significant corneal
pathology such as scarring or edema that would prevent accurate
measurement of IOP over the cornea were excluded. The risks
and benefits of participation were discussed with each participant
and informed consent was obtained. We collected patient
information on demographics, diagnosis of glaucoma, and lens
status (phakia or pseudophakia) by chart review.
Measurements
A single observer (D.S.K.) obtained all measurements. Eyes were
anesthetized with 1% proparacaine. At each time point, IOP
measurements were obtained from the central cornea and temporal sclera with the edge of the pneumatonometer probe (Model
30 Classic; Reichert Ophthalmic Instruments, Depew, NY) placed
directly temporal 1 mm from the limbus with the patient in primary gaze, which centered the probe approximately 3.5 mm
posterior to the limbus. Corneal and temporal scleral pneumotonometry measurements (abbreviated as corneal IOP and scleral
IOP, respectively) were obtained at baseline in both eyes before
injection, and then serial measurement were obtained in the
treated eye immediately after injection and 10, 20, and 30 minutes
after injection. All measurements were obtained with patients
sitting up. For each pair of measurements, we checked the corneal
IOP before the scleral IOP. All corneal measurements had a
standard deviation of less than 0.5 mmHg and all scleral measurements had a standard deviation of less than 1 mmHg for IOPs
between 0 and 40 mmHg and a standard deviation of less than 1.5
mmHg for IOPs of more than 40 mmHg. The waveform was
examined for good quality in all measurements with IOPs of less
than 40 mmHg, where it was within the limits of the paper
printout. We measured central corneal thickness by pachymetry
(DGH-550 Pachette 2; DGH Technology, Inc, Exton, PA),
averaging 5 measurements, at the time of the baseline
measurements.
correlation coefficient is not appropriate because it does not take
into account the lack of independence between repeated measurements for the same subject.6 Instead, we calculated a withinsubjects correlation coefficient, which removes the variation
between subjects to examine whether an increase in a variable
within the same subject is associated with an increase in another
variable.6 Similarly, agreement between scleral and corneal IOP
was analyzed using a Bland-Altman plot with correction for multiple measurements per subject using MedCalc Statistical Software
(MedCalc Software, Ostend, Belgium).7 The data were fit with a
linear mixed model with random slope and intercept using R (R
Foundation for Statistical Computing, Vienna, Austria).
Confidence intervals were derived from bootstrap analysis, an
iterative resampling of the data. Covariate analysis was
performed using the linear mixed model and likelihood ratio test
with P < 0.05 considered statistically significant.
Results
Thirty-three patients ranging in age from 34 to 94 years were
included in the study. Baseline characteristics are shown in
Table 1. Pseudophakia was present in 52% of patients and
glaucoma was present in 15% of patients. A total of 164 serial
paired measurements of corneal and scleral IOP were obtained in
the treated eye (1 subject missed 1 time point). Corneal IOPs
ranged from 9 to 61.5 mmHg and scleral IOPs ranged from 13.5
to 74 mmHg. Thirty-two patients had baseline measurements of
scleral and corneal pneumatonometry in the contralateral untreated
eye. At baseline, the difference between scleral and corneal
pneumatonometry measurements in the 2 eyes of individual patients was correlated significantly (r ¼ 0.75; P < 0.001; Fig 1).
We used correlation to analyze the linear association between
serial scleral and corneal IOP by pneumatonometry and found that
they were significantly correlated (r ¼ 0.94; P < 0.001) for the
injected eyes (Fig 2). The data were fit using a linear mixed model,
which takes into account longitudinal measurements over time.
This analysis resulted in the following equation: scleral IOP ¼
0.97 corneal IOP þ 10.0. The standard deviation of the
residuals, an error measurement for the entire model, was 2.78
mmHg. The slope (mean standard deviation, 0.970.21
mmHg) was statistically significant (P < 0.001) and showed a
nearly 1:1 relationship between changes in scleral and corneal
IOP on average with some variability between individual patients
(Fig 3A). Similarly, the intercept (mean standard deviation,
10.05.83 mmHg) was statistically significant (P < 0.001), but
demonstrated greater variability among patients (Fig 3B).
A Bland-Altman plot was created to examine the agreement of
scleral IOP and corneal IOP over a range of IOP using data from
Table 1. Baseline Characteristics
No. of patients enrolled
Mean age SD (yrs)
Eye (no.)
Right
Left
Mean CCT SD (mm)
Lens status (no.)
Phakic
Pseudophakic
Glaucoma (no.)
33
74.113.4
20
13
552.437.0
16
17
5 (2 POAG, 1 steroid-induced, 2 NOS)
Statistical Analysis
The Pearson correlation coefficient is reported herein. For paired
data with more than 1 time point for each study subject, an ordinary
1772
CCT ¼ central corneal thickness; NOS ¼ not otherwise specified;
POAG ¼ primary open angle glaucoma; SD ¼ standard deviation.
Kuo et al
Scleral Pneumatonometry
serial IOP measurements (Fig 4). Scleral IOP averaged 9.0 mmHg
higher than corneal IOP (95% limits of agreement, 1.5 to 19.5
mmHg). Importantly, there was not a trend toward larger
differences at higher IOPs, but there were fewer data points and
more outliers at higher values. For measurements with a mean of
scleral and corneal IOP less than 40 mmHg, which are potentially
more clinically relevant, 81.7% of the measurements were within
5 mmHg of the mean difference between scleral and corneal IOP.
To test the impact of age, eye laterality, central corneal thickness, lens status (phakia versus pseudophakia), and glaucoma on
the relationship between scleral and corneal pneumotonometry, we
added these covariates to the linear mixed model and evaluated
them with a likelihood ratio test. None of these factors was statistically significant (Table 2).
Discussion
Figure 1. Scatterplot showing baseline differences in scleral and corneal
pneumatonometry in the 2 eyes of each patient. The value for the eye
scheduled to undergo treatment with an antievascular endothelial growth
factor agent is plotted on the x-axis and for the contralateral eye on the
y-axis. Pearson correlation coefficient is shown. Reference values x ¼ y are
shown as a dotted line. IOP ¼ intraocular pressure.
Figure 2. Scatterplot showing scleral versus corneal pneumatonometry.
Data points from all measurements were plotted. The within-subjects
Pearson correlation coefficient, evaluating the relationship between
scleral and corneal intraocular pressure (IOP) in the same subject over
multiple measurements, is shown. The solid line represents line of best
fit from a linear mixed model with random slope and intercept (scleral
IOP ¼ 0.97 corneal IOP þ 10.0).
This study was designed to evaluate the relationship between serial corneal and scleral pneumatonometry over a
wide range of physiologic and pathologic levels of IOP. We
found that scleral pneumatonometry was significantly
correlated to corneal pneumatonometry, but was biased toward higher values. Although we did not compare scleral
pneumatonometry directly with Goldmann applanation, the
relationship between Goldmann applanation and corneal
pneumatonometry has been well described (Tapplanation ¼
Tpneumatonometry 1.2), and corneal pneumatonometry
has been reported to be age-independent and to correlate
best with manometric IOP measurements compared with
applanation and Tono-Pen (Reichert Ophthalmic Instruments, Depew, NY) IOP measurements in patients.8
Scleral pneumatonometry previously was found to be
increased compared with both corneal measurements and
assigned IOP, and our results are consistent with prior reports
(Table 3).4,5 The difference measured between scleral and
corneal IOP likely reflects differences in the biomechanical
properties between cornea and sclera, which can vary by
quadrant and anterioreposterior location within an individual.9,10 We chose to obtain measurements over the sclera
temporally in primary gaze because eccentric eye position can
change IOP measurement and the temporal region is the most
accessible and would be preserved even after glaucoma surgery.11,12 However, more studies are required to determine
the optimal location for scleral pneumatonometry in patients.
Our results showed a significant correlation (r ¼ 0.94)
between scleral and corneal pneumatonometry in patients
using a different approach from Kapamajian et al5 (r ¼
0.57). Kapamajian et al obtained one-time measurements
of scleral and corneal pneumatonometry in patients, which
can give variable results for the relationship between corneal
and scleral measurements because of the individual differences in scleral rigidity. In our study using patients from the
retina clinic receiving intraocular injections, we were able to
obtain multiple measurements per patient over a short
period, spanning a large range of IOPs in each subject. This
strategy allowed us to remove the variation among subjects
and evaluate whether an increase in scleral pneumatonometry was associated with an increase in corneal pneumatonometry within an individual using a within-subjects
Pearson correlation coefficient.
1773
Ophthalmology Volume 122, Number 9, September 2015
Figure 3. Estimated (A) slope and (B) intercept for individual patients from a linear mixed model with random slope and random intercept. Slope represents the change in scleral intraocular pressure (IOP) per unit of change in corneal IOP. Intercept represents a systematic difference in baseline values
between scleral and corneal IOP. The solid lines represent the means (slope, 0.97; intercept, 10.0) and the dotted lines show the 95% confidence intervals
(slope, 0.87e1.06; intercept, 7.55e12.50) derived from bootstrap analysis, using repetitive data resampling to obtain a normal distribution of values for a
collection of samples. The random effect standard deviation (SD) represents an additional variable accounting for stochastic differences between subjects. A
histogram with the distribution of values is shown on the right of each graph.
In serial IOP measurements, we found close to a 1:1 linear
relationship between changes in scleral and corneal pneumatonometry. This finding supports that following scleral
pneumatonometry measurements would be useful clinically,
because differences in scleral tonometry measurements reflect
differences in corneal tonometry even at pathologically
elevated levels of IOP. To calculate the predicted corneal IOP
from a scleral IOP measurement, our data yielded the
following equation: corneal IOP ¼ 1.04 scleral IOP 10.37. This formula is remarkably similar to the one we
found in our study of cadaveric eyes (Table 3).4 The standard
1774
deviation of the residuals, which is a measure of the accuracy
of predictions made with our model, is 2.8 mmHg. Therefore,
measured differences greater than this value are likely to
represent true changes in scleral pneumatonometry. This
value is similar to the 95% measurement accuracy
published for corneal pneumatonometry (1.5 mmHg
between 0 and 40 mmHg and 3.5 mmHg between 40 and
80 mmHg).13
On average, we found that scleral pneumatonometry was
approximately 10 mmHg higher than corneal pneumatonometry in our model. This value was close to that obtained
Kuo et al
Scleral Pneumatonometry
Table 2. Covariates Do Not Impact Relationship between Scleral
and Corneal Pneumotonometry
Figure 4. Difference plot of scleral and corneal pneumatonometry versus
the mean of scleral and corneal pneumatonometry. Agreement between
scleral and corneal intraocular pressure (IOP) was analyzed using a BlandAltman plot with correction for multiple measurements per subject. Data
from each subject are plotted with a different symbol. Perfect agreement
would show a value of 0 for the difference between scleral IOP and corneal
IOP across the range of IOPs. Mean IOP of 40 mmHg is marked as a
clinically relevant reference point. Scleral IOP showed a bias toward higher
values compared with corneal IOP, averaging 9.0 mmHg (95% limits of
agreement, 1.5 to 19.5 mmHg). SD ¼ standard deviation.
by analyzing the raw data using a Bland-Altman plot. The
Bland-Altman plot also showed that the average difference
between scleral and corneal pneumatonometry measurements appeared consistent over the range of IOPs that we
tested. Furthermore, at more clinically relevant levels of
IOP, such as mean scleral and corneal IOPs of less than 40
mmHg, more than 80% of measurements were within 5
mmHg of the average difference between scleral and corneal
IOP (Fig 4).
Ideally, to estimate best the corneal IOP from a scleral
IOP measurement, a baseline measurement of corneal and
scleral pneumatonometry should be obtained in the eye of
interest before the development of significant corneal
pathology to know the exact relationship between these
measurements. In practice, this may not be feasible, and our
data support the use of the contralateral eye to estimate
the baseline difference between scleral and corneal IOP
in the eye of interest, because the 2 eyes are significantly
Factor
P Value
Age
Eye
Central corneal thickness
Lens status
Glaucoma
0.36
0.41
0.48
0.60
0.92
correlated within an individual (Fig 1). Using this
calculated relationship between scleral and corneal
pressure in the contralateral eye of an individual would be
more accurate than using an estimate from a populationbased equation. Because the relationship between Goldmann applanation tonometry and pneumotonometry is
linear, one could use the baseline difference between
Goldmann applanation of the cornea and scleral pneumatonometry in the contralateral eye to estimate the predicted
Goldmann applanation value for the eye of interest using
scleral pneumatonometry.8
In our patient population, the relationship between scleral
and corneal pneumatonometry was not impacted by age, eye
laterality, central corneal thickness, glaucoma, or lens status.
Age has been shown to affect scleral rigidity and was reported to affect the relationship between scleral and corneal
pneumatonometry significantly in the study of Kapamajian
et al.5,14 One possibility for age not affecting our model
significantly is that we had an older population of patients.
The mean age of our population was 74.113.4 years
compared with 54.417.7 years in the study by Kapamajian
et al. Although the range of ages in our patient population
did span 34 to 94 years, we may not have been powered
adequately to detect a difference in age on scleral IOP.
Furthermore, there may be changes to the sclera related to
intravitreal injections or underlying disease that were not
assessed. Additionally, we evaluated only adult patients in
our study, which limits the generalizability of our results.
Additional studies are needed to understand the relationship
of scleral IOP to corneal IOP for children because there are
significant differences in scleral rigidity between adult and
pediatric populations.
Accurate and reliable IOP measurements are important
for both diagnosing and treating glaucoma. In patients for
whom corneal measurements are not possible or are unreliable, scleral pneumatonometry should be considered as a
potential alternative. Our results support consistency of the
Table 3. Studies of Scleral Pneumatonometry
Study
Correlation (r)
Mean Difference
Equation
Kuo et al
Lin et al4
Kapamajian et al5
0.94
d
0.57
9.0 mmHg compared with corneal IOP
13.2 mmHg compared with assigned IOP*
8.08 mmHg compared with corneal IOP
corneal IOP ¼ 1.04 scleral IOP 10.37
assigned IOP ¼ 1.01 scleral IOP 14.14
corneal IOP ¼ 0.32 scleral IOP 0.05 age þ 11.90
IOP ¼ intraocular pressure; d ¼ no data.
*Assigned IOP > corneal IOP by 3.78.
1775
Ophthalmology Volume 122, Number 9, September 2015
relationship between scleral and corneal pneumatonometry
across the range of physiologic and pathologic IOPs for
individual patients and show that changes in scleral pneumatonometry reflect changes in corneal pneumatonometry.
Acknowledgments. The authors thank Seth Blumberg, MD,
PhD, and Travis Porco, MPH, PhD, for their guidance with statistical analysis.
References
1. Talajic JC, Agoumi Y, Gagné S, et al. Prevalence, progression,
and impact of glaucoma on vision after Boston type 1 keratoprosthesis surgery. Am J Ophthalmol 2012;153:267–74.
2. Kamyar R, Weizer JS, de Paula FH, et al. Glaucoma associated
with Boston type I keratoprosthesis. Cornea 2012;31:134–9.
3. Crnej A, Paschalis EI, Salvador-Culla B, et al. Glaucoma
progression and role of glaucoma surgery in patients with
Boston keratoprosthesis. Cornea 2014;33:349–54.
4. Lin CC, Chen A, Jeng BH, et al. Scleral intraocular
pressure measurement in cadaver eyes pre- and postkeratoprosthesis implantation. Invest Ophthalmol Vis Sci
2014;55:2244–50.
5. Kapamajian MA, de la Cruz J, Hallak JA, Vajaranant TS.
Correlation between corneal and scleral pneumotonometry: an
alternative method for intraocular pressure measurement. Am J
Ophthalmol 2013;156:902–6.
6. Bland JM, Altman DG. Calculating correlation coefficients
with repeated observations: part 1dcorrelation within subjects. BMJ 1995;310:446.
7. Bland JM, Altman DG. Agreement between methods of
measurement with multiple observations per individual.
J Biopharm Stat 2007;17:571–82.
8. Eisenberg DL, Sherman BG, McKeown CA, Schuman JS.
Tonometry in adults and children. A manometric evaluation of
pneumotonometry, applanation and Tono-Pen in vitro and
in vivo. Ophthalmology 1998;105:1173–81.
9. Friberg TR, Lace JW. A comparison of the elastic properties of human choroid and sclera. Exp Eye Res 1988;47:
429–36.
10. Patel H, Gilmartin B, Cubbidge RP, Logan NS. In vivo
measurement of regional variation in anterior scleral resistance
to Schiotz indentation. Ophthalmic Physiol Opt 2011;31:
437–43.
11. Moses RA, Lurie P, Wette R. Horizontal gaze position effect
on intraocular pressure. Invest Ophthalmol Vis Sci 1982;22:
551–3.
12. Nardi M, Bartolomei MP, Romani A, Barca L. Intraocular
pressure changes in secondary positions of gaze in normal
subjects and in restrictive ocular motility disorders. Graefes
Arch Clin Exp Ophthalmol 1988;226:8–10.
13. Model 30 Pneumatonometer User’s Guide. 16030-101 Rev. F.
Depew, NY: Reichert Technologies; 2014:32.
14. Pallikaris IG, Kymionis GD, Ginis HS, et al. Ocular rigidity in
living human eyes. Invest Ophthalmol Vis Sci 2005;46:
409–14.
Footnotes and Financial Disclosures
Originally received: December 22, 2014.
Final revision: May 15, 2015.
Accepted: May 18, 2015.
Available online: July 10, 2015.
Manuscript no. 2014-2070.
1
Department of Ophthalmology, University of California, San Francisco,
San Francisco, California.
2
That Man May See, San Francisco, California (unrestricted grants to the
University of California, San Francisco, Department of Ophthalmology).
The funding organization had no role in the design or conduct of this
research.
Author Contributions:
Conception and design: Kuo, Ou, Jeng, Bhisitkul, Stewart, Duncan, Han
Department of Ophthalmology and Visual Sciences, University of
Maryland School of Medicine, Baltimore, Maryland.
Analysis and interpretation: Kuo, Han
Presented at: American Glaucoma Society Annual Meeting, February 2014,
Washington, DC.; Association for Research in Vision and Ophthalmology
Annual Meeting, May 2014, Orlando, FL; and Asia Association for
Research in Vision and Ophthalmology Annual Meeting, February 2015,
Yokohama, Japan.
Obtained funding: none
Financial Disclosure(s):
The author(s) have no proprietary or commercial interest in any materials
discussed in this article.
Supported in part by the National Eye Institute, National Institutes of
Health, Bethesda, Maryland (Core Grant for Vision Research no.:
EY02162); Research to Prevent Blindness, Inc, New York, New York; and
1776
Data collection: Kuo, Han
Overall responsibility: Kuo, Ou, Jeng, Bhisitkul, Stewart, Duncan, Han
Abbreviations and Acronyms:
Corneal IOP ¼ corneal pneumatonometry; IOP ¼ intraocular pressure;
Scleral IOP ¼ scleral pneumatonometry.
Correspondence:
Ying Han, MD, PhD, Department of Ophthalmology, University of California, San Francisco, 10 Koret Way K-323, Box 0730, San Francisco,
CA 94143. E-mail: [email protected].