Tissue classification during a stereotactic breast biopsy procedure using diffuse reflectance spectroscopy

Stereotactic Vacuum Assisted Breast Biopsy (VABB) is a minimal invasive
percutaneous biopsy technique for non-palpable breast lesions that appear as
microcalcifications on mammograms but are invisible on US. Microcalcifications
play a crucial role in early breast cancer diagnosis, the second leading cause
of cancer death among women. Approximately 50% of non-palpable breast cancers
are detected by mammography, exclusively by microcalcifications patterns.
Moreover, 80-90% of ductal carcinoma in situ (DCIS) lesions present with
microcalcifications only, without any accompanying mass lesions. Therefore,
non-palpable lesions with microcalcifications found in mammography should be
urgently evaluated.

Stereotactic VABB is a safe and accurate method for the evaluation of
suspicious microcalcifications and diagnosis of early breast cancer with a
71-100% sensitivity and 85-100% specificity range. VABB devices enable to
obtain multiple larger specimens under vacuum suction using a single insertion.
For cases where the microcalcification is spread over a large area, the needle
can be withdrawn and reinserted to perform biopsy retrieval in different areas.
Complications from VABB may include bleeding or pain during the procedure, as
well as postoperative pain, hemorrhaging, hematomas and infection. Scar
formation following VABB is observed in 16.1% of the procedures. While much
progress has been made over the past decade, there is still room for
improvement through the development of new technologies that increase accuracy,
safety, and cost-effectiveness of the procedure.

The accuracy of the stereotactic VABB depends, besides accurate positioning and
quality assurance during the procedure, on the number and volume (and thereby
needle thickness) of the specimens taken. The main disadvantage of stereotactic
VABB is the underestimation of invasive cancer diagnosis from biopsy specimens.
An overall underestimation rate from 8% till 88% for have been reported.
Atypical ductal hyperplasia (ADH) is generally considered a direct precursor
of low-grade DCIS and thus low-grade invasive ductal cancer. The histological
features of biopsy specimens containing ADH show many similarities of those
involving DCIS. Therefore, there would be a chance that a small sample of a
DCIS lesion may be interpreted as ADH by the pathologist [9]. The histological
underestimation in such cases may simply result from sampling error and are
inversely correlated with the amount of tissue excised. Not only is it
psychologically distressing for patients when breast cancer is underestimated,
but it also implies a delay in establishing a definitive diagnosis, hence,
appropriate treatment.

During the stereotactic VABB, the retrieval of calcification might suggest that
sufficient and representative tissue has been sampled from the targeted lesion.
Recently, new biopsy devices are introduced (Brevera® HOLOGIC) which make
direct specimen radiography during the procedure possible. However, if
radiologists decide to end the procedure as soon as calcifications are
retrieved, a false negative result may arise since a definite diagnosis is not
only based on specimens with calcifications. Zagouri et al. studied the
diagnostic value of specimens with and without calcifications and found that
VABB cores with microcalcifications are superior in DCIS and ADH diagnosis but
cores without microcalcifications may be more valuable for the diagnosis of the
invasive component. So during the procedure, the radiologist will have to
consider how many biopsies are needed to make the correct diagnosis which can
lead to under- of oversampling of breast tissue. Current developed techniques
for quality control during the procedure are not efficiently contributing to
this trade off so, improvement is needed to optimize the retrieval of the
invasive component.

In addition, stereotactic VABB procedures can be technically demanding due to
the nature of lesions and patient factors, which occasionally leads to failure
in retrieving the targeted lesion. The causes of retrieval failure include
significant bleeding, perception factors, targeting factors, unfavorable
calcification location, patient movement and technical factors. The failure
rate varies from up to 21%. In those cases, the procedure must be repeated
several times which leads to unnecessary tissue removal. Especially scattered
microcalcification locations or small clusters of microcalcifications within
the breast are difficult to target, as it is technically challenging to
identify the same calcification focus on the stereotactic images. In those
cases, sometimes up to 25 biopsies are collected to achieve calcification
retrieval. Not taking more specimens than necessary is essential to prevent
complications, maintain cost-effectiveness and to keep patient burden to a
minimum.

To summarize, quality control during the stereotactic VABB procedure is needed
to improve current limitations such as the underestimation of invasive
carcinoma in biopsy samples, the under- and oversampling of breast tissue and
to optimize microcalcification retrieval for challenging locations. This can be
possible by using an optical technology for real-time tissue sensing during the
stereotactic VABB procedure. To this end, we developed a tissue sensing
introducer with integrated optical fibers that enable diffuse reflectance
spectroscopy (DRS) measurements. The developed tissue sensing introducer can be
used in conjunction with a standard vacuum-assisted stereotactic biopsy (VAB)
device. DRS is a light-based technology that enables the discrimination of
tissue types based on their optical characteristics. The DRS measurements
reflect functional, biochemical and morphologic information of measured tissue
and in that way are able to discriminate tumorous tissue from healthy tissue.
The advantages of DRS are that it is non-destructive, does not require
exogenous contrast with dyes, and has the potential to be performed in
real-time. DRS technology has already been successfully evaluated in multiple
oncological domains for discriminating tumor tissue from healthy tissue with
classification accuracies of 0.86-1.00.

In our previous studies, we show that fiber-optic DRS is able to detect
invasive carcinoma (IC) and Ductal Carcinoma In Situ (DCIS) ex-vivo with high
accuracies (93-100%) without the substantial influence of patient factors such
as menopausal status. Our result mainly relies on the Near Infrared (NIR)
wavelength range, which eliminates the influence of blood on the measurements,
which is of great benefit for in-vivo measurements during stereotactic biopsy
procedures. With DRS measurements derived from ex-vivo breast specimens, a
real-time tissue classification algorithm was developed for the discrimination
of invasive carcinoma and DCIS from healthy breast tissue using the optical
introducer. Compared to this previous study, the current study is a step
towards the use case of optical spectroscopy during stereotactic biopsy
procedures. In this way, we will be able to predict tissue types during the
stereotactic biopsy procedure and assess whether the biopsy sample needs to be
collected before retrieval. With improved biopsy site selection during the
procedure the diagnostic yield will increase, the number of biopsies will be
reduced and the overall accuracy of the procedure will improve which results in
reduced complications for the patient.

This study is designed as a single-center diagnostic accuracy study. The
duration of the study is 2 years. To introduce tissue sensing during the
stereotactic VABB, we developed an optical introducer with DRS fibers that can
be integrated with the HOLOGIC eviva system, capable of rotating to align the
gap on the introducer with the needle aperture. This way, DRS measurements can
be performed for each clockwise position of the VAB needle. The optical
introducer makes it possible to determine the correct position of the VAB
device (in tumor or normal breast tissue) and to measure the retrieved tissue
within the biopsy cavity of the VAB needle before it is definitely sucked away
by VAB system. In this way, optimal real-time tissue feedback can be obtained
from the position of the VAB needle as well as of the actual biopsy tissue that
is taken. Since the DRS measurements will be acquired in a new configuration
(designed for Hologic eviva needles) and a different setup,, the algorithm for
tissue classification has to be further evaluated during the stereotactic
vacuum-assisted biopsy.

This study is therefore dedicated to validate the sensing VAB device and the
algorithm developed in the previous study in an existing clinical workflow
during vacuum-assisted biopsy under stereotactic guidance. In this study, we
will be able to test the optical introducer in a controlled manner by
histopathological confirmation of the optically measured locations. In
addition, we will collect a new dataset for DRS measurements in vivo with a
reliable histopathological gold standard. This will also provide us with the
opportunity to fine-tune the classification algorithm if there are differences
between the previously collected dataset based on ex-vivo measurements. The
goal is to achieve a real-time tissue classification algorithm for the
discrimination of tumor tissue (invasive carcinoma and DCIS) from healthy
breast tissue using the optical introducer with a sensitivity of at least 90%.

With this, we will be able to predict tissue types during the stereotactic
biopsy procedure and assess whether the biopsy sample needs to be collected
before retrieval. With improved biopsy selection during the procedure, the
number of biopsies can be reduced as well as the complications for the patient.
Most important, the diagnostic yield of the procedure will increase.

Single-center diagnostic accuracy study. The duration of the study is 2 years.

The procedure will take place under standard conditions in the department of
Radiology using the stereotactic guided HOLOGIC Affirm® Prone biopsy system and
HOLOGIC Suros vacuum console. The biopsy will be performed by experienced
radiologists familiar with VAB. We will use the HOLOGIC eviva Stereotactic
Guided Breast Biopsy Device 9Ga in combination with the tissue sensing
introducer [D2: IMDD - Tissue Sensing Introducer - Design]. The optical fibers
of the tissue sensing introducer will be connected to the DRS console [D2: IMDD
- DRS console], as described earlier. The DRS measurements will be performed
over a wavelength range of 400-1600 nm.

At the start of the procedure, a 3D mammography image (Tomosynthesis) will be
acquired, and the area of interest is targeted. Once the targeting is
completed, coordinates of the lesion within the breast are available, and the
needle position is determined. In cases with mainly microcalcifications, no
tumor lesion may be present. In those cases where no tumor lesion can be
recognized, the device will be moved through the most suspicious area, e.g. the
tissue with microcalcifications.

Next, the automatic targeting system will be activated, and the HOLOGIC eviva
breast biopsy device will be partly moved to the identified coordinates so that
the needle tip is just before the skin surface of the breast to ensure accurate
anesthetic placement. Thereafter, the needle will be manually advanced as usual
into the breast until the firing position has been reached. During the
insertion of the needle, DRS measurements will be acquired continuously at the
tip of the sensing VAB biopsy system (Figure 1 (a-c)). The measurements will be
obtained along the entire needle trajectory and will provide spectral
information from normal breast tissue and in larger lesions also from the
transition zone of the normal tissue to the lesion. The positioning of the VAB
needle with consecutive DRS measurements will end at the edge of the tumor
lesion, or in those cases with no clear tumor lesion, until the most suspicious
area.

A pair of pre-fire stereo images will be taken to determine that the area of
interest is correctly targeted. The device is then fired using the remote
firing mechanism advancing the needle 23 mm toward the target.

Once the needle is positioned in the lesion, DRS measurements will be made from
the tissue in the biopsy aperture/cavity (biopsy location) on different
locations of the aperture by sliding the optical introducer over the needle as
illustrated in Figure 1 (d-f). After the DRS measurements, the biopsy procedure
will be continued by pulling the optically measured tissue further into the
aperture with vacuum suction and subsequently advancing the tissue cutter. The
biopsied core will be transported into a tissue collection chamber at the back
of the biopsy device. Since we want a direct correlation between the sample and
the measured DRS signal, we take only a single core each time. To keep track of
the order of biopsy samples, the tissue filter basket will be replaced with a
new one after each biopsy retrieval. The procedure will be repeated for at
least 6 clock positions for every patient, or more if clinically indicated. For
each obtained core specimen, the developed classification algorithm will be
validated against histopathological examination.

It should be noticed that the whole stereotactic biopsy procedure is thus
performed according to the standard procedure, except for the DRS measurements
and biopsy samples handling.

The light used in the measurements is visual light and near infrared light.
This light, to the extent that we use it, has no negative side effects. No
adverse reactions have occurred for patient who participated in previous
studies with similar instruments. We do not expect any side effects or
complications when using this new instrument. The procedure time will increase
by 5 minutes.