7. Metodología
7.4. Instrumentos de recolección de datos y análisis
7.4.4. Encuesta de expectativas de la población objetivo
3.2.1 Standard solution preparation
A standard solution of phytoporphyrin (~95%, batch # JB04‐233; Fontier Scientific Inc. Logan UT,
USA) was prepared according to the method of Campbell et al. (2010), by dissolving the
phytoporphyrin in methanol (99.8%, Merck KGaA, Darmstadt, Germany) with the aid of
ultrasonication at 50 oC for 45 minutes. The solution was made to a final concentration of 3.549 µM
using Milli‐Q water (Merck Millipore Inc. Billerica, MA, USA). Phytoporphyrin has been shown to
adsorb on to glass surfaces (Campbell et al., 2010), and therefore the standard solution was
prepared and stored in polypropylene containers. The final solution was stored at room
temperature, in the dark, and was ultrasonicated prior to use if unused for several weeks.
Calibration curves were generated using the standard solution and control sera from cattle of
various species and age cohorts, including adult cattle as well as calves. The method of Campbell et
al. (2010) was replicated here, where different ratios of methanol and standard phytoporphyrin
solutions (5, 10, 20, 40, 80, 160, 320, 350 µL) were added to sera to give a desired concentration in a
total volume of 750 µL. Serum samples were diluted with water (1:1) before the
proteins in the solution (Campbell et al., 2010). Additionally, cuvettes were cleaned thoroughly with
water followed by methanol using cotton swabs, between each measurement to remove any
remaining phytoporphyrin.
3.2.2 Sample source
The samples were sourced from the sporidesmin challenge trial in dairy cattle described in Chapter
2. Venous blood samples were collected into vacutainers containing no anticoagulant (BD
vacutainer, Franklin Lakes NJ, USA). Samples were sent to the New Zealand Veterinary Pathology
(NZVP) Ltd, in Palmerston North, for liver panel testing and then forwarded to our group for further
testing. The liver panel test included ‐glutamyl transferase (GGT) and glutamate dehydrogenase
(GDH) activities, bilirubin, total protein, albumin and globulin measurements. Blood tubes were kept
in the dark by being wrapped in aluminium foil and/or placed in enclosed polystyrene or black plastic
containers to protect against light.
Full detail sampling procedures for the collection of blood samples for fluorescence analysis from the
facial eczema challenge are explained in Chapter 2. Samples were stored at ‐80 oC until required for
analysis.
3.2.3 Fluorescence measurements
All measurements were made with a Perkin Elmer LS‐50B luminescence spectrophotometer
(PerkinElmer Inc., Norwalk CT, USA) equipped with a red‐sensitive photomultiplier tube (R928;
Hamamatsu Photonics, Hamamatsu, Japan), using quartz semi‐micro cuvettes (0.8 ml, 4.75 x 4.75 x
33 mm, PerkinElmer Inc., Beaconsfield, Bucks, UK).
On the day of analysis sample tubes were thawed on ice. Thawed samples were centrifuged at 3000
rpm for 6 minutes, and then 200 µL of the serum supernatant was transferred to a quartz cuvette
before adding 200 µL of Milli‐Q water and 350 µL of HPLC grade methanol. The sample was mixed
by inverting several times.
For the measurement of phytoporphyrin the spectrophotometer was set to measure emissions
between 425 and 750 nm, at an excitation wavelength of 425 nm. The excitation and emission slit
widths were set at 15.0 nm and 20.0 nm, respectively, with an emission filter cut‐off of 515 nm and
scan speed of 100 nm/minute. The red‐sensitive photomultiplier tube (R928) was employed. A
standard calibration block was measured at the beginning of every sample batch to ensure
3.2.4 Data analysis
All fluorescence data, including calibration curves, were processed using OriginPro graphing
software (OriginLab Corporation, Northampton MA, USA). The following methods, based on those
of Campbell et al. (2010), were used to process the data. The peak areas were found by spline‐
fitting a baseline in the region of 575 – 750 nm. The phytoporphyrin peak is seen at 644 ± 5 nm. The
baseline was subtracted and the resulting peak was fitted with a Gaussian‐Lorentzian cross‐spectral
peak fit to measure the area of the extrapolated peak and to ensure that no potential underlying
spectral peaks were present. The area under the curve was compared to the phytoporphyrin
calibration curve and from this the concentration of the phytoporphyrin in the serum was
determined. A second order polynomial was fitted to compensate for slight deviations from linearity
at higher phytoporphyrin concentrations. At low concentrations (< 0.05 μM) of phytoporphyrin in
sera (< 0.05 μM) the reliability of the method decreased appreciably, therefore Gaussian ‐Lorentzian
areas under the peak measuring less than 2 were deemed too low to be determined accurately (limit
of detection S/N ratio > 2). Gaussian‐Lorentzian areas under the peak measuring less than 2 were
deemed to be too low to determine phytoporphyrin concentration accurately so were considered to
be noise in the spectra, and were removed from the data analysis accordingly. This was regarded as
the lower limit of the spectrophotometer sensitivity.
Biplots were produced to look for correlation patterns between phytoporphyrin, GGT, and GDH
activities. Common linear regression analyses, mixed effect models, ANOVA analyses and
Generalised Additive Models (GAM) were conducted using RStudio statistical software (Version
0.97.449, RStudio, Boston, MA, USA). An explanation of the GAM process can be found in Section
2.2.6.
Mixed effect models were used to take into account that individual cows may have differences
between each other and within themselves due to the temporal nature of the data. The models also
allowed for varying slopes and intercepts for each group (control, non‐responder, subclinical and
clinical). These models, including the common regression model, are nested, and thus were
compared using an F‐test to identify the best model. Furthermore, it was established whether the
regressions were group dependant or whether cow random effects were important. To assess the
relationships between phytoporphyrin and liver enzymes (GDH and GGT) a mixed effects linear
regression model was used. This took the form:
log
where was a random per‐cow effect, accounting for the fact that multiple observations on
per‐group slope. Thus, the model fits separate regressions for each group. In addition, a common
regression across all groups (by setting common and ) was used to determine the
overall relationship across all groups. The conditional R2 was then computed. Additionally, Cook’s
distance was inspected in residual versus leverage plots to identify any outliers. Cook’s distance
measured the effect of deleting a given observation. Samples with large residuals may strongly bias
the accuracy of the regression. Any outliers were removed and the regression re‐calculated.
3.3 RESULTS
3.3.1 Serum phytoporphyrin concentration in cows dosed with sporidesmin
Serum from clinically healthy animals (controls, non‐responders, and all animals during the control
weeks) generated either a weak fluorescence signal or no signal at all, at 644 nm. Serum from all
cows that developed clinical FE showed spectral features matching those of phytoporphyrin during
the experimental period, namely an emission band at 644 nm when excited at 425 nm. The
concentration of serum phytoporphyrin from clinically affected animals, taken after dosing, ranged
from < 0.05 to 1.33 µM, while those from subclinical cows ranged between < 0.05 and 0.18 µM, and
non‐responders between < 0.05 and 0.25 µM. In control cows, the phytoporphyrin concentrations
in serum ranged from < 0.05 to 0.12 µmol/L over the 44 day trial period. During the control weeks
for all of the sporidesmin‐dosed cattle the phytoporphyrin concentration ranged between < 0.05 and
0.16 µmol/L. One cow from the subclinical group (No. 395) presented higher than expected values
on two days during this period, while all other phytoporphyrin concentrations for this cow remained
in the normal range. These two days were considered to be outliers, with values of 0.79 and 0.57
µM on days ‐5 and ‐3, respectively. The phytoporphyrin concentrations shown on the days before
and after these measurements were 0.11 and 0.1 µM, respectively. Because of this, these values
were excluded from further analysis.
As with the traditional metric measurements reported in Chapter 2, the phytoporphyrin data were
further analysed using GAM to assess how the concentrations changed over the trial period, and
whether changes were related to groups and therefore the degree of damage caused by
sporidesmin. See Chapter 2, section 2.2.6, for details on how the model was fitted to the data. The
model fit and raw data for each animal are shown in Figure 3.2. The model showed that there is
evidence for a difference in the shape of the curve between groups (p < 0.001). Looking at the plot it
can be seen that this difference is caused by an increase in phytoporphyrin concentration in the
clinically inconsis Althoug was not
3.3.2 P
cows
All clinic was larg Day 7, b blisterin signs, de develop increase first sig measure Figure 3. with with between an obvio y affected a tent betwee h minimal, a significant.Phytoporp
cal cows exh ge (Table 3.1 but did not s ng until Day espite having ed reddenin es in phytopo ns of clinic ed. 2 Phytoporp hin‐group mo groups (p < 0 us increase af animals bega en individual a slight incre
phyrin conc
hibited eleva ). Cow 298show the fir 22. Further g one of the ng of the udd orphyrin con cal photosen hyrin concent odel fits show 0.001). This d fter dosing (D
an to increa s, with the fi ase did occu