Introduction
Medication-induced crystalluria is a well-recognized phenomenon and can result in
nephrolithiasis and/or acute kidney injury (AKI). Common etiologies such as indinavir,
acyclovir, and sulfadiazine, produce characteristic crystals easily identified on
microscopic evaluation of the urine.1, 2 A known, but less commonly considered, agent
is triamterene. A potassium-sparing diuretic typically combined with a thiazide in
the management of hypertension, triamterene-hydrochlorothiazide prescriptions totaled
10,970,464 in 2014.
3
Triamterene crystals, however, are often not appreciated on urine microscopy, and
histologically these crystals can be confused with calcium oxalate or 2,8-dihydroxyadenine
(2,8-DHA). Herein is presented a case of AKI associated with intratubular triamterene
crystallization confirmed by Raman spectroscopy.
Kidney biopsy remains the gold standard for definitive diagnosis of most etiologies
of kidney disease. A diagnosis is rendered largely from visual pattern recognition,
and this is limited by pathologist experience and the ability of current techniques
to adequately distinguish pathologic findings. Novel imaging methods are being applied
to kidney biopsy material that do not rely on operator recall, thereby permitting
definitive diagnosis of challenging material.
4
Raman spectroscopy is one such technique, which relies on intrinsic vibrational signatures
to perform label-free molecular fingerprinting.
5
Spontaneous Raman spectroscopy measures the inelastic scattering of light following
the interaction of monochromatic radiation with a sample, permitting noninvasive,
nondestructive analysis without the use of exogenous labels (Figure 1). Raman spectroscopy
of kidney tissue chemistry and its utility for diagnosis remain largely unexplored.
Figure 1
Raman spectroscopy microscope. The excitation source was a compact holographic grating-stabilized
785-nm laser diode. Dichroic mirror 1 redirected the collimated beam through the dual-axes
galvanometer mirrors. A ×60 oil-immersion objective lens focused the laser beam and
collected the backscattered light from the sample. Raman signals retraced the excitation
beam path and passed through dichroic mirror 1. The Raman signal was delivered to
the imaging spectrograph, whose spectral output was captured by a thermoelectrically
cooled charge-coupled device (CCD). A light-emitting diode (LED) was used as an illuminating
source and coupled with a complementary metal-oxide-semiconductor (CMOS) camera to
identify morphological features of the sample in the same field of view. A flip mirror
allowed switching between the Raman signal acquisition and bright field imaging.
Case Presentation and Methods
Case Presentation
A 51-year-old woman was hospitalized for AKI in the setting of cellulitis. Her history
was notable for Stage 3 chronic kidney disease (serum creatinine 1.8 mg/dl; estimated
glomerular filtration rate 37 ml/min per 1.73 m2), hypertension, and psoriasis. She
had been prescribed furosemide and triamterene-hydrochlorothiazide for hypertension,
and she had experienced several prior episodes of AKI attributed to pre-renal azotemia.
She had no history of nephrolithiasis, and family history was remarkable for a sister
with end-stage renal disease (ESRD). At presentation, serum creatinine was 3.2 mg/dl,
urine protein 119 mg per day, and urinalysis revealed 23 white blood cells per high-power
field without hematuria or cellular casts (Table 1). Ultrasound demonstrated the right
kidney to be 8.5 cm and left 9.2 cm, both with normal echogenicity. With antibiotics
and discontinuation of diuretics, creatinine returned to baseline. She underwent kidney
biopsy for further evaluation of her chronic kidney disease given the family history
of ESRD.
Table 1
Laboratory values
Test
Admission
2 mo before admission
Blood urea nitrogen, mg/dl
28
32
Creatinine, mg/dl
3.2
1.8
Albumin, g/dl
3.1
3.8
Urinalysis
Specific gravity
1.012
1.012
pH
6.5
6.5
Protein
Trace
Negative
Hemoglobin
Negative
Negative
Nitrite
Negative
Negative
Leukocyte esterase
Moderate
Moderate
White blood cells, hpf
23
3
Red blood cells, hpf
0
0
Epithelial cells, hpf
1
0
24-hour urine, mg
119
–
hpf, high-power field.
On biopsy, 31 glomeruli were present, of which 2 were globally sclerosed. Moderate
tubular injury was present with sparse interstitial inflammatory infiltrate and less
than 10% interstitial fibrosis. Mildly dilated tubules contained radially arranged
rod- or rhomboid-shaped crystals that filled the tubular lumen in most places, at
times associated with giant cell reaction (Figure 2a–e). The crystals were yellow-brown
on hematoxylin-eosin and periodic acid-Schiff stain, black/gray on methenamine silver,
pale blue on Masson trichrome stain, and were polarizable with Maltese cross formation.
In comparison, 2,8-DHA crystals are similar in color and shape to triamterene and
are also birefringent. Oxalate, however, is usually clear to blue and does not form
Maltese crosses.
Figure 2
Intratubular triamterene crystals (arrows). (a) Brownish-yellow with associated giant
cell reaction by hematoxylin-eosin, original magnification ×100; (b) brownish-yellow
by periodic acid–Schiff, original magnification ×100; (c) black/gray by methenamine
silver, original magnification ×100; (d) pale blue by Masson trichrome, original magnification
×160; and (e) Birefringence and Maltese cross formation under polarized light, original
magnification ×100.
Given the remarkable visual similarity to 2,8-DHA crystals seen in adenine phosphoribosyltransferase
(APRT) deficiency, APRT was sequenced and enzymatic activity was measured. No sequence
variants were detected, and enzymatic activity was within normal limits. Given the
family history of ESRD, definitive identification of the crystals to exclude APRT
deficiency was made by Raman spectroscopy.
Control regions (Figure 3a) and 3 separate biopsy sites containing crystals (Figure 3b)
were probed. The spectral similarity between the tissue crystal and that of a pure
triamterene sample is evident visibly and statistically (R = 0.75; Figure 3c). Figure 3d
shows the spectral overlay from the tissue crystals and a pure 2,8-DHA sample (R =
0.11). Additionally, the correlation coefficients of a pure sample of triamterene
with 2,8-DHA and calcium oxalate monohydrate were 0.09 and −0.06, respectively.
Figure 3
(a) Raman spectrum from tissue without crystals. (b) Raman spectra of intratubular
triamterene crystals from 3 sites. The presence of prominent peaks at 700, 1003, 1352,
and 1603 cm−1, and their relative peak intensities, were closely mirrored at all 3
crystal sites.(c) Overlay of Raman spectra of intratubular triamterene crystals (dotted)
and prepared triamterene sample (solid). (d) Overlay of Raman spectra of intratubular
triamterene crystals (dotted) and prepared 2,8-DHA sample (solid).
At 2-year follow-up, creatinine was 1.2 mg/dl (estimated glomerular filtration rate
57 ml/min per 1.73 m2) off triamterene-hydrochlorothiazide.
Methods
Five-micrometer sections of optimal cutting temperature embedded frozen tissue, triamterene
(Sigma-Aldrich, St. Louis, MO), and 2,8-DHA (Toronto Research Chemicals, Toronto,
Ontario, Canada) were placed separately on quartz coverslips. For Raman measurements,
an inverted confocal Raman microscope was adapted from a published design.
6
The excitation source was a compact LM series volume holographic grating-stabilized
laser diode (λem = 785 nm) (Ondax, Monrovia, CA) with a clean-up filter (LL01-785-12.5;
Semrock, Rochester, NY). The laser was redirected to the dual-axes galvanometer mirrors
(GVS112; Thorlabs, Newton, NJ) via a dichroic laser-flat beam splitter (LPD01-785RU-25;
Semrock). The galvanometer mirrors enable high-speed lateral scanning in the sample
plane. A ×60 oil-immersion objective lens focused the laser and collected backscattered
light from the sample. The Raman scattering light was collected by a 50-mm multimode
fiber (M14L01; Thorlabs), delivered to a HoloSpec f/1.8 spectrograph (Kaiser Optical
Systems, Ann Arbor, MI), and finally detected by an iDus charge-coupled device camera
(DU420A-BEX2-DD; Andor, Belfast, UK). Exposure time was 10 seconds at 5-mW laser power.
Cosmic ray and fluorescence signals were removed before spectral analysis. Customized
LabView 2013 (National Instruments, Austin, TX) and MATLAB 2013 (Mathworks, Natick,
MA) modules were used for system control and data analysis. Spectra were compared
by Pearson correlation coefficient.
Coding regions and introns within 20 base pairs of the exon/intron boundaries of APRT
(NM_000485.2) were amplified and sequenced in the forward and reverse directions using
automated fluorescent dideoxy sequencing methods (Baylor Miraca Genetics Laboratory,
Houston, TX). Semi-quantitative assay of APRT enzymatic activity in dried blood spots
was performed by UCSD Biochemical Genetics Laboratory (San Diego, CA).
Discussion
Combination triamterene-hydrochlorothiazide was the number 1 prescribed medication
in the United States in 1986, and as of 2014 was number 75.
3
Renal excretion represents the major route of elimination, with urinary crystals present
in 50% to 100% of individuals taking the drug.7, 8 Triamterene nephrolithiasis was
first reported in 1979 with an incidence of 1:1500 to 1:2500.9, 10 Although crystalluria
and nephrolithiasis are not rare, triamterene-associated AKI has been reported only
on occasion.
Triamterene can crystallize in the renal tubules of rats.
8
In 1986, polarizable crystals primarily within the cytoplasm of distal tubular epithelial
cells were noted in a patient following a triamterene-hydrochlorothiazide overdose
associated with AKI.
11
This was followed by a case of irreversible AKI in which yellowish, polarizable crystals
were found to obstruct the tubular lumen in association with multinucleated giant
cells.
12
Our patient’s kidney function slowly improved with discontinuation of triamterene,
leading one to speculate that the removal of precipitated triamterene from the renal
tubules may occur over time. It was not until 2014 when another 2 cases of triamterene-associated
AKI were reported, one of which was originally misdiagnosed as 2,8-DHA crystal deposition
due to APRT deficiency.
13
Our crystals were yellow-brown, polarizable with Maltese cross formation, and had
accompanying giant cell reaction, as has been reported for triamterene. In comparison,
2,8-DHA crystals are also birefringent, can be similar in color and shape to triamterene,
and may form Maltese crosses in urine but not typically in renal parenchyma.14, 15
Calcium oxalate crystals are birefringent, but are usually clear to blue and do not
form Maltese crosses.
APRT deficiency (OMIM #102600) is an autosomal recessive disorder resulting in metabolism
of adenine to 2,8-DHA. Patients are at increased risk for AKI, chronic kidney disease
is present in approximately one-third of patients, and 10% may reach ESRD. Nephrolithiasis
is common in APRT deficiency, and the stones are radiolucent.14, 16 A diagnosis of
APRT deficiency was not supported by either genetic or biochemical analysis. Given,
however, the visual similarity to 2,8-DHA crystals and the family history of ESRD,
APRT deficiency was definitively excluded by positively identifying the crystals as
triamterene by Raman spectroscopy.
The characterization of urinary calculi is commonplace, but limited work has been
done to identify crystal species in kidney tissue.17, 18, 19 Raman spectroscopy has
been used to identify oxalate in tissue from patients with known oxalosis, as well
as calcium crystals in kidney tumors.20, 21 Several groups have used Fourier-transformed
infrared spectroscopy to confirm 2,8-DHA, as well as other crystals in kidney biopsies.15,
22, 23 Our case extends the application of Raman spectroscopy to making a de novo
clinical diagnosis from the analysis of unidentified crystalline material. These results
underscore the molecular specificity of Raman spectroscopy and its ability to identify
crystals in a biopsy specimen. As databases of acquired spectra grow, Raman spectroscopy
could be used in the characterization of other biopsy findings, such as unidentified
pigmented material, constitution of tubular casts, cellular infiltrates, and drug-induced
kidney injury.
To this end, investigators have used Raman spectroscopy to identify hydroxyethyl starch
in osmotic lesions from patients with AKI after exposure to the agent.
24
The presence of hydroxyethyl starch in kidney allografts taken from hydroxyethyl starch–resuscitated
donors also has been correlated with graft outcome.
25
In a mouse model of anti-glomerular basement membrane disease, Raman spectroscopy
was able to distinguish the severity of glomerulonephritis with 98% accuracy by using
principal component and linear discriminant analysis.
26
An automated measurement of tissue injury can add significantly to the pathological
information derived from an experienced pathologist. Furthermore, Raman spectroscopy
has shown promise in prognosticating renal allograft outcome from the urine of deceased
donors.
27
Using silver nanoparticles for surface-enhanced Raman spectroscopy of deceased donor
urine, Chi et al.
27
achieved a sensitivity of 91% for identifying donor kidneys with acute tubular necrosis
but ultimately good allograft function, as compared with allografts that ultimately
displayed delayed graft function. In addition, in a preliminary study of renal allograft
recipients, silver nanoparticle-enhanced spectra obtained from urine samples collected
1 day after surgery identified allografts at risk for future dysfunction.
28
The authors postulated free heme may be the source of the prognostic Raman peak in
these samples.
As Raman spectroscopy is inexpensive and widely available in most academic environments,
there exists considerable opportunity for its application to clinical diagnosis. Its
exquisite molecular specificity, lack of sample preparation needs, and relatively
simple instrumentation allow Raman spectroscopy to serve as a powerful adjunct to
current histopathological techniques. Moreover, subtle variations in Raman spectra
can be leveraged to recognize cell types and the tissue microenvironment, as well
as to differentiate between various morphologically similar but biochemically distinct
pathologies.29, 30 Raman has low sensitivity to water, and because handheld portable
Raman spectrometers are commercially available, Raman spectroscopy is also well suited
to the analysis of urine. Future work should explore the relationship between urine
and kidney biopsy spectroscopy in crystalline nephropathies, with a possible goal
of rendering a noninvasive diagnosis.
Conclusion
In summary, triamterene-associated crystalline nephropathy is a rare but potentially
underrecognized clinical entity (Table 2). Clinicians should consider triamterene
in the differential diagnosis of AKI and nephrolithiasis. Specific morphologic features
can aid in the identification of intratubular triamterene, but significant overlap
can exist with 2,8-DHA and oxalate. The growing application of nontraditional technologies
like Raman spectroscopy to the interpretation of kidney biopsy material highlights
the value of extracting novel information from collected material.
Table 2
Teaching points
Triamterene crystallization should be considered in the differential diagnosis of
acute kidney injury.
Intratubular triamterene crystals are yellow-brown on hematoxylin-eosin and periodic
acid-Schiff, black/gray on methenamine silver, pale blue on Masson trichrome stain,
and under polarized light are birefringent with occasional Maltese cross formation.
Raman spectroscopy is a nondestructive, readily available imaging modality that can
ascertain novel information from the kidney biopsy.
Disclosure
All the authors declared no competing interests.
Author Contributions
CJS conceived the study, interpreted the data, drafted the manuscript, and provided
intellectual content to the conduct of the work. CJS had full access to the data in
the study and takes final responsibility for the decision to submit for publication.
CZ performed Raman spectroscopy, interpreted the data, revised the manuscript, provided
intellectual content to the conduct of the work, and approved the final draft. MD
performed renal histology, revised the manuscript, provided intellectual content to
the conduct of the work, and approved the final draft. RG performed renal histology,
revised the manuscript, provided intellectual content to the conduct of the work,
and approved the final draft. SB performed renal histology, revised the manuscript,
provided intellectual content to the conduct of the work, and approved the final draft.
IB performed Raman spectroscopy, interpreted the data, revised the manuscript, provided
intellectual content to the conduct of the work, and approved the final draft.