A validation study of whole slide imaging for primary diagnosis of lymphoma

Abstract

Whole slide imaging (WSI) is being increasingly used worldwide. Although previous studies have asserted the validity of WSI diagnosis, they have primarily targeted only small specimens and excluded cases requiring immunohistochemistry or special staining, such as lymphoma. The purpose of this study was to evaluate the accuracy of WSI diagnosis of lymphoma, for which 240 biopsies and resections of lymphoma cases were selected from the study set of lymphomas. All slides including H&E, immunohistochemical and special staining were digitized using a WSI image scanner. An experienced pathologist performed the WSI diagnoses, which were compared with original diagnoses based on light microscopic examinations. Discrepancy between the two interpretations were classified into three categories: concordance, minor discrepancy (no clinical significance), and major discrepancy (with clinical significance). Overall concordance between the light microscopic and WSI diagnosis was found in 223 cases (92.92%; 95%CI = 88.90–95.82), minor discrepancy in fifteen (6.25%; 95%CI = 3.54–10.10), and major discrepancy in two (0.83%; 95%CI = 0.10–2.98). Diagnosis of lymphoma using WSI appeared to be mostly accurate, suggesting that WSI may be a reliable technology for the diagnosis of lymphoma.

Full text

Pathology International. 2019;69:341–349. wileyonlinelibrary.com/journal/pin © 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
|
341
Received: 17 January 2019
|
Accepted: 21 April 2019
DOI: 10.1111/pin.12808
ORIGINAL ARTICLE
A validation study of whole slide imaging for primary
diagnosis of lymphoma
Saiful Amin
|
Taro Mori
|
Tomoo Itoh
Department of Diagnostic Pathology, Kobe University Hospital, Kobe, Japan
Correspondence:
Tomoo Itoh, MD, Department of Diagnostic
Pathology, Kobe University Hospital, 7‐5‐2,
Kusunoki‐cho, Chuo‐ku, Kobe 650‐0017,
Japan.
Email: tomitoh@med.kobe-u.ac.jp
Whole slide imaging (WSI) is being increasingly used worldwide. Although
previous studies have asserted the validity of WSI diagnosis, they have primarily
targeted only small specimens and excluded cases requiring immunohistochem-
istry or special staining, such as lymphoma. The purpose of this study was to
evaluate the accuracy of WSI diagnosis of lymphoma, for which 240 biopsies and
resections of lymphoma cases were selected from the study set of lymphomas.
All slides including H&E, immunohistochemical and special staining were
digitized using a WSI image scanner. An experienced pathologist performed
the WSI diagnoses, which were compared with original diagnoses based on light
microscopic examinations. Discrepancy between the two interpretations were
classified into three categories: concordance, minor discrepancy (no clinical
significance), and major discrepancy (with clinical significance). Overall
concordance between the light microscopic and WSI diagnosis was found in
223 cases (92.92%; 95%CI =88.90–95.82), minor discrepancy in fifteen (6.25%;
95%CI =3.54–10.10), and major discrepancy in two (0.83%; 95%CI =
0.10–2.98). Diagnosis of lymphoma using WSI appeared to be mostly accurate,
suggesting that WSI may be a reliable technology for the diagnosis of lymphoma.
KEYWORDS
diagnosis, lymphoma, validation studies, whole slide imaging
INTRODUCTION
Whole slide imaging (WSI), also known as virtual microscopy,
is the transformation of glass slides of the tissue section to
digital images by WSI scanners which are run with computer
workstations and specific software.
1–7
WSI has various
pragmatic uses such as diagnosis, education, research, and
telepathology consultation.
1–4,8–14
Another dynamic use of WSI
results in promotion of the workflow by avoiding the extra‐work
of slide allocation.
1,7,15
The swift increase of improvements in
this field and their various advantages may lead to a change
from light microscopy to digital pathology for routine
pathological diagnosis.
2,6
At present there are many WSI
scanners, related workstations and specific software available
that are capable of performing good quality digitalization.
1,3
Recently several reports have indicated that WSI can be
used for routine pathological diagnosis.
12,16,17
The College
of American Pathologists (CAP) proposed in May 2013, 13
graft guidelines for conducting validation studies of WSI for
clinical use as well as use with digital pathology devices for
primary diagnosis.
1,3,9,12,13,16,18
Several validation studies
for WSI were conducted more than a decade ago with many
of them suggesting that WSI diagnosis is not inferior to light
microscopic interpretation.
2,4,7,10,12,14,16–22
However, many

of them used only a small number of specimens alone and
cases requiring immunohistochemistry or special staining,
such as lymphoma cases were excluded from the studies.
WSI validation studies for lymphoma have not been yet
published and more studies are required prove the clinical
values of WSI for pathological diagnosis.
2,4,7,14,18,22–29
In this study, we assessed the usefulness of WSI in
comparison with that of light microscopy (LM) for the
diagnosis of lymphoma.
MATERIALS AND METHODS
For this study, 240 biopsies and resections of serial cases were
retrieved from a study set of lymphoid lesions. All original
diagnoses had been established by one of the authors who is an
experienced pathologist (T.I., hereafter referred to as “the
evaluator”) based on the routine microscopic examinations, at
least a half year before this study. More than half of the cases
were nodal lesions (54.59%), while extra‐nodal lesions consisted
of those of the gastrointestinal tract (19.59%), skin (3.75%), and
others (Table 1).
All slides of the 240 cases had already been scanned for
educational purposes with a scanner Nanozoomer 2.0 RS
(Hamamatsu Photonics, Hamamatsu, Japan) at a magnification
of ×20 except bone marrow lesions because we thought the
image quality of WSI was insufficient and focusing problems with
bone marrow evaluation occurred relatively frequently. All slides
of H&E, immunohistochemical, and special staining were
scanned properly according to CAP guidelines.
9
The images
of scanned slides were reviewed through a high‐resolution
monitor (Iiyama Prolite B2888UHSU; Iiyama Corporation,
Iiyama, Japan). The scanned images were stored in a mass
storage environment. A viewer NDP.view2 (Hamamatsu Photo-
nics) was used for evaluation on WSI. Table 2 shows details of
the 240 cases.
For simulating the actual WSI diagnostic environment, two
authors (S.A., T.M.) played the role of “preparer”. They
provided clinical information that had already been recorded
with WSI images to the evaluator who was blinded to the
original diagnosis. The evaluator then made his diagnoses
using this clinical information and WSI images as well as
results obtained with H&E and immunohistochemistry. The
original‐and the WSI‐based diagnoses were compared for
each case and the degree of agreement was determined by
discussions with all evaluators. The results thus obtained
were categorized into three classes: (i) concordance, defined
as complete agreement between the two diagnoses; (ii)
minor discrepancy, slightly different without any clinical or
prognostics significances; (iii) major discrepancy, different
interpretations with significant clinical implications. For cases
a with discrepant diagnosis, those cases were reviewed by
the evaluators, after which the final diagnosis was reached
and determined on whether discrepancy occurred on WSI
due to image quality or human factors.
The percentages of agreement between microscopic and
WSI‐based diagnoses were calculated, and 95% confidence
intervals were also determined by using the Clopper‐Pearson
exact method for binomial distributions. Concordant rates were
calculated for the entire cohort as well as for individual histology
types. SAS software (9.4 version; SAS Institute Inc, Cary, NC,
USA) was used for statistical analyses.
RESULTS
Diagnostic agreement between WSI and microscopic
diagnoses
The overall extent of agreement between WSI and microscopic
diagnoses is shown in Table 3. Of 240 cases examined, the
diagnosis was concordant for 223 cases (92.92%; 95%
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Table 1 Distribution of studied specimens
Location of specimens Number of cases Percentage (%)
Lymph node 131 54.59
Gastrointestinal tract 47 19.59
Skin 9 3.75
Mediastinum 8 3.33
Lung 7 2.93
Breast 6 2.50
Testis 5 2.08
Brain 4 1.67
Pharynx 4 1.67
Adrenal gland 3 1.25
Nasal cavity 2 0.83
Orbit 2 0.83
Tonsil 2 0.83
Thyroid 2 0.83
Liver 2 0.83
Urinary bladder 2 0.83
Ovary 2 0.83
Uterine cervix 2 0.83
240 100%
Table 2 Summary of cases and scanned slides
Cases 240
Type of specimens Biopsy and resection
Number of biopsy cases 201
Number of resected cases 39
Total number of slides 1560
Average slides per case 6.5
342 S. Amin et al.

confidence interval, 88.90–95.82). Minor discrepancy was
detected in 15 cases, while major discrepancy was identified
in the remaining two cases.
Similarly, the concordant rate was calculated for individual
diagnostic categories and the most common diagnosis was
diffuse large B‐cell lymphoma (96/240 cases, 40.0%),
followed by follicular lymphoma (41 cases, 17.08%),
nonspecific reactive lymphadenopathy (13 cases, 5.41%)
and others (Table 4). Cases with minor discrepancy
consisted of 11/41 cases of follicular lymphoma, 1/4 cases
of high‐grade B‐cell lymphoma, not otherwise specified
(NOS); 1/6 cases of atypical lymphoid cell infiltration, 1/7
cases of classical Hodgkin lymphoma, and 1/5 cases of
nodular lymphocyte predominant Hodgkin lymphoma. The
two major discrepant cases were one case of nodular
lymphocyte predominant Hodgkin's lymphoma and one case
of tuberculosis concomitant with low‐grade B‐cell lymphoma.
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Table 3 Degree of agreement between WSI and LM‐based
diagnosis among overall cases
Degree of
agreement
Number of
cases Percentage
95%
Confidence
interval
Concordance 223 92.92 88.90–95.82
Minor
discrepancy
15 6.25 3.54–10.10
Major
discrepancy
2 0.83 0.10–2.98
Table 4 Summary and degree of agreement between WSI and LM‐based diagnosis among diagnostic subtype of malignant lymphoma
Diagnosis of cases
Number of
cases
Percentage (%)
of cases
Concordance:
percentage; 95%CI
Minor discrepancy:
percentage; 95%CI
Major discrepancy:
Percentage: 95%CI
Diffuse large B‐cell lymphoma 96 40.00 100; 96.2–100 0; 0–3.8 0; 0–3.8
Follicular lymphoma 41 17.08 73.2; 57.1–85.8 26.8; 14.2–42.9 0; 0–8.6
Extra‐nodal Marginal Zone
Lymphoma of mucosa associated
lymphoid tissue (MALT lymphoma)
9 3.75 100; 66.4–100 0; 0–33.6 0; 0–33.6
Mantle cell lymphoma 7 2.93 100; 59.0–100 0; 0–41.0 0; 0–41.0
Low‐grade B‐cell lymphoma 5 2.08 100; 47.8–100 0; 0–52.2 0; 0–52.2
High‐grade B‐cell lymphoma, NOS 4 1.67 75.0; 19.4–99.4 25.0; 0.6–80.6 0; 0–60.2
Burkitt lymphoma 2 0.83 100; 15.8–100 0; 0–84.2 0; 0–84.2
SLL/ CLL 3 1.25 100; 29.2–100 0; 0–70.8 0; 0–70.8
Lymphoplasmacytic lymphoma 2 0.83 100; 15.8–100 0; 0–84.2 0; 0–84.2
Plasma cell myeloma 2 0.83 100; 15.8–100 0; 0–84.2 0; 0–84.2
Angioimmunoblastic T‐cell
lymphoma
5 2.08 100; 47.8–100 0; 0–52.2 0; 0–52.2
Peripheral T‐cell lymphoma, NOS 9 3.75 100; 66.4–100 0; 0–33.6 0; 0–33.6
NK/T‐cell lymphoma 4 1.67 100; 39.8–100 0; 0–60.2 0; 0–60.2
Anaplastic large cell lymphoma 5 2.08 100; 47.8–100 0; 0–52.2 0; 0–52.2
Adult T‐cell lymphoma 2 0.83 100;15.8–100 0; 0–84.2 0; 0–84.2
T‐lymphoblastic lymphoma 2 0.83 100;15.8–100 0; 0–84.2 0; 0–84.2
Classic Hodgkin lymphoma 7 2.93 85.7; 42.1–99.6 14.3; 0.4–57.9 0; 0–41.0
Nodular sclerosis Hodgkin
lymphoma
4 1.67 100; 39.8–100 0; 0–60.2 0; 0–60.2
Mixed cellularity classic Hodgkin
lymphoma
4 1.67 100; 39.8–100 0; 0–60.2 0; 0–60.2
Lymphocyte rich classic Hodgkin
lymphoma
2 0.83 100;15.8–100 0; 0–84.2 0; 0–84.2
Nodular lymphocyte predominant
Hodgkin lymphoma
5 2.08 60.0; 14.7–94.7 20.0; 0.5–71.6 20.0; 0.5–71.6
Atypical lymphocytic infiltration 6 2.50 83.3; 35.9–99.6 16.7; 0.4–64.1 0; 0–45.9
Nonspecific reactive
lymphadenopathy
13 5.41 100; 75.3–100 0; 0–24.7 0; 0–24.7
Tuberculosis with atypical B‐cell
infiltrate
1 0.42 0; 0–97.5 0; 0–97.5 100; 2.5–100
Abbreviations: CI, confidence interval; CLL, chronic lymphocytic leukemia; NK, natural killer cell; NOS, not otherwise specified; SLL, small
lymphocytic lymphoma.
Validation of WSI for lymphoma diagnosis 343

Review of the discrepant cases
One case with major discrepancy was an inguinal lymph
node biopsy. The original light microscopic diagnosis was
nodular lymphocyte predominant Hodgkin lymphoma, while
evaluation with WSI, resulted in a diagnosis of peripheral
T‐cell lymphoma, NOS. Figure 1 shows representative
pictures of this case, which show a vague nodular pattern
and a configuration mainly consisting of small T‐cells with a
definite atypia: irregular shape and size. On WSI, it was
difficult to determine whether these T‐cells were neoplastic,
resulting in a final determination as neoplastic but without
confidence. In addition, lymphocyte predominant (LP) cells
were a little less striking on WSI than on LM, so that the
evaluator could not detect the cells. Histology review of this
case confirmed that the original diagnosis was more
appropriate than WSI diagnosis. In this case, image quality
partly affected diagnostic accuracy because the details of the
nuclear shape were hard to see on WSI. However human
factors were considered to be also related to this discre-
pancy.
Another case with major discrepancy was a cervical lymph
node biopsy from a right cervical lymphadenopathy. The
original light microscopic diagnosis was tuberculosis with
atypical B‐cells infiltrate. In this case, large foci of caseous
necrosis were found surrounded by epithelioid cells with
Langerhans’s giant cells. Background lymphocytes were
small and uniform with bland‐looking morphology, but
showed diffuse positivity for CD20. The reason for the
original microscopic diagnosis was the presence of a few
acid‐fast bacilli on Ziehl‐Neelsen staining, but these were not
detected by WSI at ×20 scanning magnification. In contrast,
WSI diagnosis was low‐grade B‐cell lymphoma, NOS with
granulomatous reaction with undetermined significance.
After the histology review, a final diagnosis of tuberculosis
with coexisting low‐grade B‐cell lymphoma, NOS was made.
Figure 2 shows the histological features of this case.
Of the 15 cases with minor discrepancies, 11 were related
to grading of follicular lymphoma. Eight of these cases were
diagnosed as grade 2 on WSI, while the original diagnosis
was grade 1. Another two cases, originally rated as grade
3a, were diagnosed as grade 2 on WSI and the last one was
diagnosed as diffuse large B‐cell lymphoma on WSI, while
the original LM diagnosis was follicular lymphoma grade 3b.
However, grading of follicular lymphoma and differentiating
between diffuse large B‐cell lymphoma and follicular
lymphoma, grade 3b are considered to be fairly subjective,
and inter‐and intra‐observer variations are common even for
microscopic examinations, not from image quality of WSI,
and do not affect the selection of therapies and are therefore
not a significant problem. These differences of the remaining
four cases with minor discrepancies were considered to be
the result of human factors, and not of image quality. One of
these cases with an original diagnosis of high‐grade B‐cell
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Figure 1 First major discrepant case, diagnosed as peripheral T‐cell lymphoma, NOS on WSI, but original diagnosis nodular lymphocyte pre-
dominant Hodgkin lymphoma. (a) H&E staining showing LP cells (arrow) and surrounded by atypical small to medium‐sized lymphocytes. (b)CD20
immunohistochemistry showing positivity to LP cells (arrows) surrounded by T‐cell rosettes. (c) CD3 immunohistochemistry showing irregularity of
shape and size of T‐cells, which misled to a wrong diagnosis as peripheral T‐cell lymphoma.
344 S. Amin et al.

lymphoma, NOS was diagnosed as diffuse large B‐cell
lymphoma on WSI. The lymphoma was composed of
medium to large B‐cells with an unusual immunophenotype,
partial cyclin D1‐positive, CD23 strongly positive, CD5‐
positive (weak), and a very high MIB1 index. Although a
pleomorphic variant of mantle cell lymphoma or high‐grade
transformation of chronic lymphocytic leukemia/small lym-
phocytic lymphoma were suspected, no specific diagnostic
name could be assigned, also due to the small size of the
specimen and absence of previous history of lymphoma.
Another of these four cases, originally diagnosed as “atypical
lymphoid cell infiltration (indeterminate for malignancy)”was
diagnosed on WSI as “probable T‐cell rich B‐cell lymphoma”.
The discrepancies in both diagnoses were inconclusive,
rated as in the area of “borderline”, and categorized as
minor. In one case of classical Hodgkin lymphoma, NOS was
diagnosed as nodular sclerosis Hodgkin lymphoma on WSI;
in the last case, nodular lymphocyte predominant Hodgkin
lymphoma with a T‐cell rich B‐cell lymphoma‐like area was
diagnosed as simply T‐cell rich B‐cell lymphoma on WSI.
Generally, the distinction between nodular lymphocyte
predominant Hodgkin lymphoma and T‐cell rich B‐cell
lymphoma is sometimes not very clear, so that this case
was also classified as a minor discrepancy.
DISCUSSION
Our results were compared with those of previously validated
and published studies on various subspecialties, such as
skin,
8,11,19,28
breast,
5,6,22,23
prostate,
24,29
liver,
7
urinary
bladder,
25,26
gastrointestinal,
14,20
gynaecological,
2,10
pae-
diatric,
4,27
and general pathology
12,16,17
. A summary of some
of these previously validated study results is shown Table 5.
Ninety‐six cases (40.0%) in this study were diffuse large
B‐cell lymphoma, with a concordance rate of 100%. This rate
can be attributed to the fact that the cellular atypia in diffuse
large B‐cell lymphoma is almost always striking and never
overlooked even on WSI. Diffuse large B‐cell lymphomas
rarely show an unusual morphology, such as epithelioid or
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Figure 2 Second major discrepant case, diagnosed as low‐grade B‐cell lymphoma but original diagnosis was tuberculosis. (a) H&E showed
wide‐spreading epithelioid granulomas with caseous necrosis with diffuse infiltrate of small lymphocyte in background. (b) CD20 im-
munohistochemistry showed total diffuse dense infiltrate of B‐cell in background, suggesting low‐grade B‐cell lymphoma.
Validation of WSI for lymphoma diagnosis 345

sarcomatoid. Such unusual cases were not included in this
study, which can be considered one of the limitations of our
study. However, we suggest WSI is still useful for such
cases, because we always use immunohistochemistry for
such difficult cases and never reach final conclusion based
only on H&E morphology. To confirm this hypothesis, a
larger study with more cases is necessary.
Two cases of Burkitt lymphoma were included in this study
with a concordance rate of 100%. The morphological
diagnosis of high‐grade B‐cell lymphoma was easily made,
and immunohistochemical results for CD10, bcl‐2 and MIB1,
which are useful markers for differentiation of this type of
lymphoma from diffuse large B‐cell lymphoma with similar
morphology, could be accurately evaluated with WSI.
For B‐cell lymphomas composed of small to medium‐sized
lymphocytes, such as follicular lymphoma, mantle cell
lymphoma, small lymphocytic lymphoma, low‐grade
B‐cell lymphoma, lymphoplasmacytic lymphoma, and extra‐
nodal marginal zone B‐cell lymphoma of MALT, an antibody
panel consisting of CD20, CD5, CD10, CD23, cyclin D1, and
bcl‐2, was very helpful for the diagnosis. For follicular
lymphomas, a combination of CD10 and bcl‐2 was also
found to be useful. The use of these panels, made the
diagnosis less difficult for the evaluator, except for follicular
lymphoma grading, while the morphology of small lympho-
cytes was somewhat difficult to see in detail on WSI. The
appropriate use of an antibody panel could compensate for
the resolution problem, so that WSI can be considered
suitable for these types of lymphomas.
As for T or NK/T‐cell lymphomas, the evaluator did not find
it difficult to diagnose typical cases on WSI, such as
peripheral T‐cell lymphoma, NOS, angioimmunoblastic T‐
cell lymphoma, anaplastic large cell lymphoma, adult T‐cell
lymphoma, T‐lymphoblastic lymphoma, and NK/T‐cell lym-
phoma, although details of the cell morphology were some-
what difficult to see. T‐cell lymphoma usually show overt
cellular atypia: convoluted shapes, notches, or grooves, and
may display characteristic pale cytoplasm and extensive
proliferation of high endothelial venules, especially in
angioimmunoblastic T‐cell lymphomas, all of which were
easily observed on WSI. CD3 immunohistochemistry is
useful for the detection of the nuclear atypia, which can
also be evaluated on WSI with almost complete accuracy in
typical cases. For anaplastic large cell lymphomas, the
combination of CD30 and ALK immunohistochemistry
proved to be convenient and did not present the evaluator
with any difficulties for interpretation on WSI. However, it was
difficult to differentiate between T‐cell lymphoma and a
reactive condition in cases with small atypical T‐cell
infiltrates. T‐cells may generally show mild cellular atypia
even in cases with a reactive condition. So, we concluded
almost all discrepancies were not from the quality of images,
but one exception was T‐cell morphology. The concordance
rate was high in all T‐cell lymphoma (100%). However, the
evaluator sometimes had to make diagnoses without
absolute confidence and one nodular lymphocyte predomi-
nant Hodgkin lymphoma was curiously diagnosed as a
peripheral T‐cell lymphoma, NOS. We therefore suggest
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Table 5 Summary of previously published study
Author Study field
Number of cases
or specimens Concordance rate
Minor
discrepancy
rate
Major
discrepancy
rate
Ordi et al.
2
Gynaecological specimens 452 94.2% 3.8% 2%
Tabat et al.
16
General pthology 900 cases,
1070 specimens
95.6% 3.6% 0.8%
Al‐Janabi
et al.
20
Gastrointestinal tract 100 cases 95% 5% None
Al‐Janabi et al.
5
Breast 100 93% 6% 1%
Al‐Janabi
et al.
19
Dermatopathology 100 94% 6% None
Al‐Janabi et al.
4
Paediatric pathology 100 90% 8% 2%
Al‐Janabi
et al.
26
Urinary system 100 87% 8% 5%
Saco et al.
7
Liver 176 Overall intra‐
observer 96.6%
for observer 1
and 90.3% for
observer 2
No description No description
Fallon et al.
10
Ovary 52 96% 4%
Our study Malignant lymphoma 240 92.92% 6.25% 0.83%
346 S. Amin et al.

WSI may not be entirely suitable for T‐cell lymphomas, and
recommend to changing over to LM if there is any concern
about the accuracy of the diagnosis based on WSI. For
confirming this result, it is necessary to perform case study
with larger number of cases and “evaluators”.
For classical Hodgkin lymphomas, no major discrepancies
were noted, but only a few cases with minor discrepancies
(14.3%) as mentioned above. For subtypes of classic
Hodgkin lymphoma such as mixed cellularity classic Hodgkin
lymphoma, nodular sclerosis classic Hodgkin lymphoma and
lymphocyte rich classic Hodgkin lymphoma, concordance
between WSI‐based and LM‐based diagnosis were almost
same, and no major or minor discrepancies were noted. It
was not hard to find Hodgkin/Reed‐Sternberg cells on WSI,
and immunohistochemical findings obtained with CD30,
CD15, Pax‐5 etc. could be accurately evaluated on WSI
without any difficulty. But for nodular lymphocyte predomi-
nant Hodgkin lymphomas we detected one major discre-
pancy and one minor discrepancy out of a total of five cases
as previously mentioned, the overall concordance rate was
not satisfactory, so that the availability of only five cases of
nodular lymphocyte predominant Hodgkin lymphoma is
considered one of this study’s limitations.
As for reactive changes in lymph nodes, the concordance
rate of 100% is almost identical for both WSI and LM
evaluations. In our study, however, only thirteen cases
(5.41%) comprising one dermatopathic lymphadenopathy, one
Kimura disease, one Castleman’s disease and ten nonspecific
reactive lymphadenopathies showed reactive conditions,
making this another limitation of our study. Generally speaking,
it is sometimes very difficult to distinguish neoplastic from
reactive lymphocytes under inflammatory conditions such as
small gastric ulcer bed tissues. Since this study is based on a
study set of “lymphoma”cases, these 13 cases were not
included, and an evaluation of the accuracy of WSI diagnoses
for such cases requires a larger‐scale case study.
Optical magnification is considered to be an important
factor for performing WSI effectively. We could not use ×40
scanning for this study because we were using a previously
constructed study set already scanned at ×20. It is therefore
unknown how the results could have improved if we had
used ×40. In our study, however, the evaluator did not have
much difficulty with evaluating few cases at ×20 magnifica-
tion. In lymphomas composed of large cells, such as diffuse
large B‐cell lymphomas, cell morphology was easily ob-
served with ×20, while even in low‐grade B‐cell lymphomas,
an accurate diagnosis could be achieved by evaluating the
immunoarchitectures of CD20, CD10, CD5, CD23, and
cyclin D1 etc., for which the results could be easily
interpreted with ×20. In contrast, the evaluator found it
difficult to determine whether small irregular‐shaped T‐cells
were neoplastic, which led to a major discrepancy in one
case. Although, we do not claim that ×20 is entirely sufficient
for lymphoma diagnosis using WSI, we believe that WSI
even of cases with ×20 magnification presents almost no
practical problems for lymphoma diagnosis, provided LM‐
backup can be used for difficult cases. In addition, it is
estimated that ×40 scanning is going to be the mainstream
procedure with future increases in storage so that magnifica-
tion problems will become less significant.
Another potential limitation of our study is that the same
pathologist diagnosed all cases by using LM and after more
than a half year rediagnosed all cases on WSI, thus
precluding intra‐observer variation. Since our results were
based on the diagnoses of only one pathologist, it might be
considered a limitation of the study. However, it was
impossible to increase the number of evaluators, because
sufficient wash‐out time between the original LM and WSI
diagnoses had to be established. In addition, the number of
cases of each type of lymphoma were too small in this study.
It is hoped further studies will validate our results.
Excluding bone marrow cases is another one of our study’s
limitations. When we prepared the study set, bone marrow
cases were not scanned, except for informative or problematic
ones, because we thought the image quality of WSI was
insufficient and focusing problems with bone marrow evalua-
tion occurred relatively frequently. As a result, the remaining
number of bone marrow cases was considered to be too small
for evaluation. Another study is also required to confirm the
validation of WSI for bone marrow biopsy.
In this study, molecular testing findings were not used for
either original or WSI diagnoses, which were all made with
clinical information and morphological findings alone. In actual
practice, we make pleotropic diagnoses on the basis of
molecular testing combined with the findings of flow‐cytometry
genetic and chromosome analyses. However, if we had used
molecular testing, the results might have been better. For this
reason, as for conventional LM diagnoses, we recommend
using molecular testing as well for difficult cases.
Another bias of this study is that all immune‐stained and
special stained slides, in addition to HE slides, are provided
from the beginning, this might give some help to make a
diagnosis. For example, when the evaluator inclined to make
a diagnosis of follicular lymphoma, the subsequent slides
including cyclin‐D1 positivity may change his mind and leads
to the correct diagnosis of mantle cell lymphoma. Although it
was very difficult to avoid this bias, we thought this didn’t
have a vast impact on the result, because we always apply a
set of markers depending on the morphology. For example,
lymphomas composed with small to medium‐sized lympho-
cytes, we almost always add panel of antibodies; CD5,
CD10, CD23, LEF1, and cyclin D1.
An advantage of WSI is that many slides can simultaneously
be viewed on one monitor at the same time, which is very
helpful for comparing results obtained with H&E and immuno-
histochemistry, or with different kinds of antibodies. In addition,
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Validation of WSI for lymphoma diagnosis 347

WSI is especially useful for evaluation of aberrant expressions
produced by particular markers such as CD5, CD43 or light
chain restriction. Another advantage is its immediacy under
remote diagnostic circumstances. Lymphoma is one of the most
complex and difficult fields in diagnostic pathology, resulting in
the pathologist tending to experience difficulty and stress with
routine lymphoma diagnosis. If a telepathology system using
WSI can be employed and a consultation system connecting
with experts can be established, the quality of the diagnosis is
bound to be much higher, and the implementation of WSI
technology in the field of lymphoma diagnosis may effectively
play an important role in the improvement of the quality and
safety of healthcare. In addition, WSI makes it possible to
render image analysis and may eventually be considered as
computer‐aided diagnostic technology that will help to lessen
intra‐or inter‐observer variability and enhance the ability to
perform diagnoses based on objective evaluation.
The findings of our study show high concordance and lower
discordance rates for WSI and LM evaluations, and suggest
that WSI can be used as an effective diagnostic tool for
primary diagnosis of lymphoma cases. However, we found
WSI image quality was still insufficient for evaluating detailed
cellular morphology, especially in making a distinction on
whether T‐cell were neoplastic or not. We recommend to
changing over to LM if there is a lack of confidence in reaching
a diagnosis or if diagnosing a lymphoma proves to be difficult,
and if necessary, that molecular testing should be used in
conjunction with conventional LM lymphoma diagnosis.
ACKNOWLEDGMENTS
We gratefully acknowledge the all kinds of logistical and
technical supports from the Department of Diagnostic
Pathology, Kobe University Hospital, Japan.
DISCLOSURE STATEMENT
None declared.
AUTHOR CONTRIBUTIONS
TI and SA: Conception or design of the work. TI: Data
collection. SA, TI and TM: Data analysis and interpretation.
SA: Drafting the article. TI: Critical revision of the article. SA,
TM and TI: Final approval of the version to be published.
REFERENCES
1 Snead DRJ, Tsang YW, Meskiri A et al. Validation of digital
pathology imaging for primary histopathological diagnosis.
Histopathology 2016; 68: 1063–72.
2 Ordi J, Castillo P, Saco A et al. Validation of whole slide imaging
in the primary diagnosis of gynaecological pathology in a
University Hospital. J Clin Pathol 2015; 68:33–39.
3 Pantanowitz L, Valenstein PN, Evans AJ et al. Review of the
current state of whole slide imaging in pathology. J Pathol Inform
2011; 2:2.
4Al‐Janabi S, Huisman A, Nikkels PG, Ten Kate FJ, Van Diest PJ.
Whole slide images for primary diagnostics of paediatric
pathology specimens: A feasibility study. J Clin Pathol 2013;
66: 218–23.
5Al‐Janabi S, Huisman A, Willems SM, Van Diest PJ. Digital slide
images for primary diagnostics in breast pathology: A feasibility
study. Hum Pathol 2012; 43: 2318–25.
6 Elmore JG, Longton GM, Pepe MS et al. A randomized study
comparing digital imaging to traditional glass slide microscopy
for breast biopsy and cancer diagnosis. J Pathol Inform 2017;
8: 12.
7 Saco A, Diaz A, Hernandez M et al. Validation of whole‐slide
imaging in the primary diagnosis of liver biopsies in a University
Hospital. Dig Liver Dis 2017; 49: 1240–46.
8 Brick KE, Comfere NI, Broeren MD, Gibson LE, Wieland CN.
The application of virtual microscopy in a dermatopathology
educational setting: Assessment of attitudes among dermato-
pathologists. Int J Dermatol 2014; 53: 224–27.
9 Pantanowitz L, Sinard JH, Henricks WH et al. Validating whole
slide imaging for diagnostic purposes in pathology: Guideline from
the college of American pathologists pathology and laboratory
quality center. Arch Pathol Lab Med 2013; 137:1710–22.
10 Fallon MA, Wilbur DC, Prasad M. Ovarian frozen section
diagnosis: Use of whole‐slide imaging shows excellent correla-
tion between virtual slide and original interpretations in a large
series of cases. Arch Pathol Lab Med 2010; 134: 1020–23.
11 Velez N, Jukic D, Ho J. Evaluation of 2 whole‐slide imaging
applications in dermatopathology. Hum Pathol 2008; 39:1341–9.
12 Bauer TW, Schoenfield L, Slaw RJ, Yerian L, Sun Z, Henricks
WH. Validation of whole slide imaging for primary diagnosis in
surgical pathology. Arch Pathol Lab Med 2013; 137: 518–24.
13 Thrall MJ, Wimmer JL, Schwartz MR. Validation of multiple
whole slide imaging scanners based on the guideline from the
College of American Pathologists Pathology and Laboratory
Quality Center. Arch Pathol Lab Med 2015; 139: 656–64.
14 Singson RP, Natarajan S, Greenson JK, Marchevsky AM. Virtual
microscopy and the Internet as telepathology consultation tools:
A study of gastrointestinal biopsy specimens. Am J Clin Pathol
1999; 111: 792–95.
15 Gilbertson JR, Ho J, Anthony L, Jukic DM, Yagi Y, Parwani AV.
Primary histologic diagnosis using automated whole slide
imaging: A validation study. BMC Clin Pathol 2006; 6:4.
16 Tabata K, Mori I, Sasaki T et al. Whole‐slide imaging at primary
pathological diagnosis: Validation of whole‐slide imaging‐based
primary pathological diagnosis at twelve Japanese academic
institutes. Pathol Int 2017; 67: 547–54.
17 Campbell WS, Lele SM, West WW, Lazenby AJ, Smith LM,
Hinrichs SH. Concordance between whole‐slide imaging and
light microscopy for routine surgical pathology. Hum Pathol
2012; 43: 1739–44.
18 Weinstein RS, Graham AR, Richter LC et al. Overview of
telepathology, virtual microscopy, and whole slide imaging:
Prospects for the future. Hum Pathol 2009; 40: 1057–69.
19 Al‐Janabi S, Huisman A, Vink A et al. Whole slide images for
primary diagnostics in dermatopathology: A feasibility study.
J Clin Pathol 2012; 65: 152–58.
20 Al‐Janabi S, Huisman A, Vink A et al. Whole slide images for
primary diagnostics of gastrointestinal tract pathology: A
feasibility study. Hum Pathol 2012; 43: 702–07.
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
348 S. Amin et al.

21 Ozluk Y, Blanco PL, Mengel M, Solez K, Halloran PF, Sis B.
Superiority of virtual microscopy versus light microscopy in
transplantation pathology. Clin Transplant 2012; 26: 336–44.
22 Reyes C, Ikpatt OF, Nadji M, Cote RJ. Intra‐observer reprodu-
cibility of whole slide imaging for the primary diagnosis of breast
needle biopsies. J Pathol Inform 2014; 5:5.
23 Krishnamurthy S, Mathews K, McClure S et al. Multi‐institutional
comparison of whole slide digital imaging and optical micro-
scopy for interpretation of hematoxylin‐eosin–stained breast
tissue sections. Arch Pathol Lab Med 2013; 137: 1733–39.
24 Rodriguez‐Urrego PA, Cronin AM, Al‐Ahmadie HA et al. Inter-
observer and intraobserver reproducibility in digital and routine
microscopic assessment of prostate needle biopsies. Hum
Pathol 2011; 42:68–74.
25 Compérat E, Egevad L, Lopez‐Beltran A et al. An interobserver
reproducibility study on invasiveness of bladder cancer using
virtual microscopy and heatmaps. Histopathology 2013; 63:
756–66.
26 Al‐Janabi S, Huisman A, Jonges GN, ten Kate FJ, Gold-
schmeding R, van Diest PJ. Whole slide images for primary
diagnostics of urinary system pathology: A feasibility study.
J Renal Inj Prev 2014; 3:91–96.
27 Arnold MA, Chenever E, Baker PB et al. The College of American
Pathologists guidelines for whole slide imaging validation are
feasible for pediatric pathology: A pediatric pathology practice
experience. Pediatr Dev Pathol 2015; 18:109–16.
28 Al Habeeb A, Evans A, Ghazarian D. Virtual microscopy using
whole‐slide imaging as an enabler for teledermatopathology:
A paired consultant validation study. J Pathol Inform 2012; 3:3.
29 Camparo P, Egevad L, Algaba F et al. Utility of whole slide
imaging and virtual microscopy in prostate pathology. APMIS
2012; 120: 298–304.
© 2019 Japanese Society of Pathology and John Wiley & Sons Australia, Ltd
Validation of WSI for lymphoma diagnosis 349