Merck Sharp & Dohme Corp. v. Amneal Pharmaceuticals LLC

Filing 178

OPINION. Signed by Judge Sue L. Robinson on 1/30/2017. (nmfn)

IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF DELAWARE

MERCK SHARP & DOHME CORP.,
Plaintiff,
v.

)
)
)
)
)

Civ. No. 15-250-SLR

)

AMNEAL PHARMACEUTICALS LLC,
Defendant.

)
)
)

Jack B. Blumenfeld, Esquire and Derek J. Fahnestock, Esquire of Morris, Nichols, Arsht
& Tunnell LLP, Wilmington, Delaware. Counsel for Plaintiffs. Of Counsel: Nicolas
Barzoukas, Esquire, Joshua Davis, Esquire, Kevin E. Cadwell, Esquire, and Lisa M.
Thomas, Esquire of Reed Smith LLP.
Anne Shea Gaza, Esquire, Robert M. Vrana, Esquire, and Samantha G. Wilson,
Esquire of Young Conaway Stargatt & Taylor, LLP, Wilmington, Delaware. Counsel for
Defendant. Of Counsel: Constance S. Huttner, Esquire, Michael H. lmbacuan, Esquire,
Stephanie J. Kamerow, Esquire, Caroline Sun, Esquire of Budd Larner, P.C.

OPINION

Dated: January ..'bO, 2017
Wilmington, Delaware

I. INTRODUCTION
This action arises out of the filing of Abbreviated New Drug Application ("ANDA")
No. 207989 by defendant Amneal Pharmaceuticals LLC ("Amneal") seeking to produce
and market a generic mometasone furoate nasal spray. (D.I. 56) On March 20, 2015,
plaintiff Merck Sharp & Dohme Corp. ("Merck") brought this action alleging infringement
of U.S. Patent No. 6,127,353 ("the '353 patent"). 1 (D.I. 1) Merck filed an amended
complaint on September 4, 2015, 2 which Amneal answered on September 18, 2015.
(D.I. 56; D.I. 59) The court held a Markman hearing on July 31, 2015 and issued a
claim construction order on September 3, 2015, construing certain disputed limitations.
(D.I. 1073) The court held a final pretrial conference on June 7, 2016, and a two-day
bench trial on June 21 and 22, 2016 on the infringement issue. The parties have since
completed post-trial briefing. The 30-month stay of FDA final approval on Amneal's
ANDA expires on August 4, 2017. (D.I. 142, ex. 1 at~ 46) The court has jurisdiction
over this matter pursuant to 28 U.S.C. §§ 1331, 1338(a), and 1400(b). Having
considered the documentary evidence and testimony, the court makes the following
findings of fact and conclusions of law pursuant to Federal Rule of Civil Procedure
52(a).

1

The '353 patent is listed in the Food and Drug Administration's ("FDA's") publication
titled "Approved Drug Products with Therapeutic Equivalence Evaluations" (known as
the "Orange Book") for Nasonex® ("Nasonex"). (D.I. 56 at~ 15)
2

3

Agreed to by stipulation (D.I. 51) and so ordered by the court.

The parties agreed to be bound by the claim construction and invalidity decisions in
the co-pending litigation Merck Sharp & Dahme Corp. v. Teva Pharmaceuticals USA,
Inc., Civ. No. 14-874-SLR (D. Del.). The court concluded that the asserted claims of the
'353 patent were valid. (Civ. No. 14-874, D.I. 133, 204, 205)

II. FINDINGS OF FACT AND CONCLUSIONS OF LAW

A. Technology at Issue
1. Development of MFM
Anhydrous mometasone furoate ("MFA") was first synthesized and patented by a
Merck chemist, Dr. Elliot Shapiro, in the early 1980s. (D.I. 163 at 4) After MFA was
discovered, its unique physical properties that prevented it from dissolving in water or
known pharmaceutically acceptable compounds kept it on the "backburner" for further
research. (Id.) Years later, scientists found that MFA dissolved in a new
pharmaceutical solvent and developed MFA for the treatment of psoriasis, a skin
condition. (Id.)
In the late 1980s, a formulator at Merck, Dr. Yuen, led a project seeking to
develop mometasone furoate for nasal applications. As a result of this project,
mometasone furoate monohydrate ("MFM") was developed. MFM has the chemical
name,

9a,21-dichloro-16a-methyl-1,4-pregnadiene-11~.1 ?a-diol-3,20-dione-17-(2'-

furoate) monohydrate and the following chemical structure:
0

0

. .,,,>--()
II

Cl

H

()

(D.I. 163 at 46; '353 patent)
MFA and MFM are polymorphs. MFM differs from MFA in that every molecule of
MFM is associated with a molecule of water, whereas no water is present in the crystal

3

lattice structure of MFA. The difference between the molecular structures of MFM and
MFA causes changes to the solid structure of the two crystalline forms. (0.1. 163 at 5;
PTX 184 )

2. Development of Nasonex
Upon discovering MFM, Dr. Yuen determined that using MFM as a suspension in
water with other excipients provided a stable formulation. (0.1. 163 at 6) The formation
was further developed and ultimately was approved as Nasonex. The formulation is
protected by the '353 patent. (Id.)
Nasonex is indicated for the treatment of perennial allergenic rhinitis, seasonal
allergic rhinitis, nasal polyps, and congestion associated with the nasal symptoms of
allergic rhinitis (0.1. 142, ex. 1 at~ 26) The product insert for Nasonex states:
"[Nasonex] Nasal Spray 50 mcg is a corticosteroid demonstrating potent antiinflammatory properties." (Id.

at~

35) It further states: "The precise mechanism of

corticosteroid action on allergic rhinitis is not known. Corticosteroids have been shown
to have a wide range of effects on multiple cell types ... and mediators ... involved in
inflammation." (Id.) Nasonex contains MFM as its active pharmaceutical ingredient
("API"). (Id. at ,-r 34)

3. The '353 patent
The '353 patent, titled "Mometasone furoate monohydrate, process for making
same and pharmaceutical compositions," issued on October 3, 2000. (JTX 1) Merck
alleges infringement of independent claims 1 and 6 and dependent claims 9-12. (0.1.

4

Xioming Chen et al., Solid State Characterization of Mometasone Furoate Anhydrous
and Monohydrate Forms, 94 Journal of Pharmaceutics Sciences (2005) ("the Chen
article"). (PTX 18)

4

142, ex. 1 at~ 15) The patent claims MFM, a process for preparing MFM by
crystallization from a saturated aqueous water miscible organic solution, and aqueous
stable pharmaceutical compositions of MFM. ('353 patent, 1:31-48) Independent claim
1 recites "9a,21-dichloro-16a-methyl-1,4-pregnadiene-11r3.17a-diol-3,20-dione-17-(2'furoate) monohydrate" and independent claim 6 recites "[a] pharmaceutical composition
comprising mometasone furoate monohydrate in a carrier consisting essentially of
water."

4. The accused ANDA product
Amneal's ANDA product is a generic mometasone furoate nasal spray, 50 mcg,
using MFA as the active pharmaceutical ingredient. Amneal's ANDA product has a
proposed shelf-life of two years. Merck does not allege that the pre-formulation active
pharmaceutical ingredient used in Amneal's ANDA product contains MFM or otherwise
infringes the '353 patent. (D.I. 142, ex. 1 at~~ 43-47; D.I. 163 at 3; PTX 23)

B. Infringement Standard
A patent is infringed when a person "without authority makes, uses or sells any
patented invention, within the United States ... during the term of the patent." 35
U.S.C. § 271 (a). To prove direct infringement, the patentee must establish that one or
more claims of the patent read on the accused device literally or under the doctrine of
equivalents. See Advanced Cardiovascular Sys., Inc. v. Scimed Life Sys., Inc., 261
F.3d 1329, 1336 (Fed. Cir. 2001). A two-step analysis is employed in making an
infringement determination. See Markman v. Westview Instruments, Inc., 52 F.3d 967,
976 (Fed. Cir. 1995), aff'd, 517 U.S. 370 (1996). First, the court must construe the
asserted claims to ascertain their meaning and scope, a question of law. See id. at

5

976-77; see also Teva Pharms. USA, Inc. v. Sandoz, Inc.,_ U.S._, 135 S. Ct. 831, 837
(2015). The trier of fact must then compare the properly construed claims with the
accused infringing product. See Markman, 52 F.3d at 976. This second step is a
question of fact. Spectrum Pharm., Inc. v. Sandoz Inc., 802 F.3d 1326, 1337 (Fed. Cir.
2015) (citing Bai v. L & L Wings, Inc., 160 F.3d 1350, 1353 (Fed. Cir. 1998)).
"Direct infringement requires a party to perform each and every step or element of a
claimed method or product." Exergen Corp. v. Wal-Mart Stores, Inc., 575 F.3d 1312,
1320 (Fed. Cir. 2009) (quoting BMC Res., Inc. v. Paymentech, L.P., 498 F.3d 1373,
1378 (Fed. Cir. 2007)). "If any claim limitation is absent ... , there is no literal
infringement as a matter of law." Bayer AG v. Elan Pharm. Research Corp., 212 F.3d
1241, 1247 (Fed. Cir. 2000). If an accused product does not infringe an independent
claim, it also does not infringe any claim depending thereon. Ferring B. V. v. Watson

Labs., lnc.-Florida, 764 F.3d 1401, 1411 (Fed. Cir. 2014) (citing Wahpeton Canvas Co.,
Inc. v. Frontier, Inc., 870 F.2d 1546, 1552 (Fed. Cir. 1989) ("One who does not infringe
an independent claim cannot infringe a claim dependent on (and thus containing all the
limitations of) that claim.")). However, "[o]ne may infringe an independent claim and not
infringe a claim dependent on that claim." Monsanto Co. v. Syngenta Seeds, Inc., 503
F.3d 1352, 1359 (Fed. Cir. 2007) (quoting Wahpeton Canvas, 870 F.2d at 1552)
(internal quotations omitted). The patent owner has the burden of proving literal
infringement by a preponderance of the evidence. Octane Fitness, LLC v. ICON Health
& Fitness, Inc.,_ U.S._, 134 S. Ct. 1749, 1758 (2014).

C. Analysis

6

The question for infringement is whether Amneal's ANDA product (an aqueous
suspension made with prior art MFA) contains any patented MFM during the product's
two-year shelf life. 5

1. Samples
Amneal produced samples of Batch Nos. BB-ST-13003A (manufactured October
28, 2013), 13005A (manufactured November 22, 2013), and 13006A (manufactured
December 9, 2013) (collectively "the Exhibit Batches") and samples of Batch No. RD3965-162 ("the R&D Batch") to Merck. 6 (JTX 8-10, 14-16) The samples were packaged
in nasal spray bottles, Amneal's finished product form. (D.I. 176 at 55:17-23, 110:23111 :5) Amneal produced samples from Batch No. RD-3965-157 ("the Commercial
Batch"), after which Amneal changed its manufacturing process. The court ruled that
evidence of MFM in such samples could not be used to show infringement.7 (D.I. 89,

5

Merck contends that the MFA in Amneal's product will eventually convert to MFM and
the only question is "when" the conversion will occur. (D.I. 163 at 45) Amneal disputes
both "if' and "when" the MFA in its ANDA product will convert to MFM, arguing in part
that even Merck's expert, Dr. Adam Matzger ("Dr. Matzger"), opined that there are some
MFA formulations that will not convert. (D.I. 165 at 29-30; D.I. 176 at 125:22-126:1)
The parties spend considerable energy discussing the evidence put forth regarding
conversion and disputing the issue of "if" conversion will occur. The court only
addresses the arguments material to the infringement question.
6

Merck tested the R&D Batch, but did not analyze the results, because it determined
that the R&D batch was not relevant to infringement. (D.I. 176 at 87:10-12)
7

The court allowed Dr. Matzger to answer a few questions regarding the Commercial
Batch, explaining that such information might be appropriate if it pertained to the
likelihood of conversion (from MFA to MFM) in the sample environment. Dr. Matzger
testified that he identified MFM crystals in the Commercial Batch using both Raman
spectroscopy and single crystal XRay diffraction ("SCXRD"), the gold standard for
polymorphic identification. He opined that he was able to perform SCXRD, as "the
crystals had grown for about a year-and-a-half, so they were larger." (D.I. 176 at 67:221, 68:25-69:5, 69:22-70:9) None of the elicited testimony is helpful to the question at
bar.

7

98, 154) Amneal produced samples of another commercial-sized batch, Batch No. BBST-16001("Batch16001") in glass bottles. (D.I. 130; D.I. 176at110:20-111:5; JTX 5,
13) The samples produced were drawn after manufacture on January 11, 2016 ("Batch
16001 Day 1"). Three days later, on January 14, 2016, the batch was mixed and a
certain portion was removed and packaged in nasal spray bottles ("Batch 16001A").
The next day, the batch was mixed, samples were drawn ("Batch 16001 Day 4"), and
the remainder was packaged in nasal spray bottles ("Batch 16001A"). 8 (JTX 11-12)
Amneal stipulated that the Batch 16001 Day 1 samples produced to Merck are
representative of its ANDA product. (D.I. 130)

2. Merck's thermodynamic stability study
Merck's expert, Dr. Matzger, conducted a thermodynamic stability test "to
establish the thermodynamic stability of the monohydrate relative to the anhydrous
form." He added an amount of MFM equal to the amount of MFA in one of the sample
bottles from the Exhibit Batches and subjected the bottle to vigorous shaking at 500
RPM. After 27 days, all of the MFA had converted to MFM. He explained that shaking
"increase[s] mass transport" to help the conversion from the "less stable form ... to the
more stable form" and "break up the viscosity" of the suspension. Dr. Matzger
concluded that "the monohydrate is the more stable form in the environment of
Amneal's formulation." He testified that the study established that "conversion will
occur," but not "when it will occur." He further explained that he "intentionally added
[MFM] so that the conversion could take place with both forms present, and so [he]

8

The court adopts the parties' nomenclature of referring to the ANDA product packaged
in nasal-spray bottles on consecutive days as Batch 16001A. (D.I. 163 at 8-10)

8

wouldn't know if [MFM] would become present or when it would become present if [he]
hadn't added it;" a person of skill in the art would "need to know all of those things to say
what the rate would be in the proposed ANDA product."
(D.I. 176 at 60-63, 114:19-23) Amneal's expert, Dr. Robin Rogers ("Dr. Rogers"), does
not dispute the findings of the study, opining that the addition of MFM "changed the rate
of nucleation" and the kinetics, such that the formulated product "now had a crystal form
that it preferred." (D.I. 177 at 427-429)
3. Batch 16001
Merck's expert, Dr. Bernhardt Trout ("Dr. Trout"), explained that polymorphic
conversion is governed by the "basic principles of thermodynamics and chemical
kinetics." Further, polymorphic conversion involves nucleation, which "is the first step in
which the smallest crystal of the new polymorph is formed. And then the second step ..
. is growth, where that crystal then grows to a larger crystal." As polymorphic
conversion is affected by energy and collisions, "the more one mixes, the more
vigorously something is mixed, the more energy [and] the more collisions are
imparted."9 (D.I. 176 at 197-199, 208:16-209:5) An unstable system tends to convert to
a stable system. (Id. at 222:25-223:8) Dr. Trout opined that it is "very difficult to make

9

Referencing Chong-Hui Gu, Victor Young, Jr., and David Grant, Polymorph Screening:
Influence of Solvents on the Rate of Solvent-Mediated Polymorphic Transformation, 90
Journal of Pharmaceutical Sciences 11, (2001) ("the Gu article"). (DTX 180) Dr. Trout
explained that the Gu article is an example of "one solid crystalline form transforming to
another and the effects of agitation" thereon. (D.I. 176 at 258:2-7) Dr. Trout admitted
that the Gu article did not specifically study MFA or MFM and focused on solvent
mediated transformation of a chemical called SMZ. Moreover, the mixing was done
with a wrist shaker. (D.I. 176 at 232-233) The court concludes that the Gu article is not
helpful to the issue at hand, i.e.,whether mixing would facilitate the conversion of MFA
to MFM in the aqueous pharmaceutical suspension at bar.
9

predictions" on how mixing would affect a specific system, and "to verify in a given
sample whether there was conversion, it needs to be tested empirically." (Id. at 199:1217, 207:11-18, 236:11-237:21) Applying these principles to Batch 16001, he concluded
that "[m]ore vigorous mixing enhances the likelihood of nucleation. And once nucleation
has occurred, for example, as Dr. Matzger said in the [Batch] 16001 Day 1 sample, the
additional mixing, the vigor of the mixing, the collision imparted [and] the energy
imparted by this industrial mixing would increase the likelihood of growth of [MFM] in the
nucleated system." (Id. at 208: 14-209:3) He also concluded that "as a whole, the vigor
of the mixing of the 16001 Batch was greater than that of the Exhibit Batches."
Moreover, the concentration of the MFA varied within the manufacturing process, which
might also affect the conversion. (Id. at 201-204, 265:22-268:3; PTX 97, 98)
Dr. Matzger testified that "the way you mix something [and] the vigor with which
you mix something affects the polymorphic conversion." (D.I. 176 at 108:8-16)
Therefore, he "would expect to see at least as much" MFM in the later samples of Batch
16001 as in the Batch 16001 Day 1 sample, and would expect that "the amount would
have increased," because "the additional mixing, if anything, would be expected to
increase conversion." 10 (Id. at 111 :20-112:23) Dr. Matzger admitted that he did not
investigate a rate of conversion in response to mixing speed, and stated that the
conversion is a solution mediated process. (Id. at 135-137)
Dr. Rogers disagreed with Dr. Trout's opinions, testifying that the conversion of
MFA to MFM is difficult, as there is a high energy barrier. He explained that in a

10

The actual mixing speeds and times for the various samples are redacted. (JTX 7-10;
DTX 26)
10

solution, conversion is a three-part process - dissolution of the crystal form, nucleation,
and crystal growth of the new form. He opined that MFA "has very good physical
stability under humidity." 11 (D.I. 177 at 417-420) He described that the Exhibit Batches
and the 16001 Batch "were subjected to the highest energy stirring" for the same
amount of time. Therefore, if the high energy could cause conversion, there should
have been MFM present in the Exhibit Batches. 12 As to the 16001 Batch, he opined
that Dr. Trout's "cumulative mixing" theory fails to take into account the periods of rest in
between the later mixing steps. (Id. at 433-440, 444:23-445: 17) He agreed that
concentration can be related to how fast a material crystalizes and that "theoretically, it
is possible" that "increasing mixing or agitation[] sometimes can result in an increase of
the nucleation rate of a crystal." (Id. at 465-466)
The parties dispute whether Amneal should have provided samples from Batch
16001 Day 4 and Batch 16001A ("additional samples") to Merck. 13 Amneal asserts that
the additional samples would be cumulative to those provided (Batch 16001 Day 1 and
the Exhibit Batches). Merck requests that the court conclude that the additional
samples would have contained MFM because of the additional mixing. From the expert
testimony, the court concludes that generally additional (or faster) mixing tends to
promote conversion of MFA to MFM. Neither party, however, has offered a

11

Referencing the Chen article. (PTX 18) Dr. Trout criticized reliance on the Chen
article because it studied the physical stability under humidity, which is completely
different than an aqueous suspension. (D.I. 176 at 250-51)
12

Dr. Rogers also opined that agitation can increase or decrease the rate of nucleation
in a particular system. J.W. Mullin, Crystallization: Chapter 5, Nucleation, 191 (41h ed.
2001 ). (DTX 84) Dr. Trout criticized the reliance on this reference, opining that it
focuses on nucleation in a solution not in an aqueous suspension. (D.I. 177 at 492-93)
13

This has been an ongoing dispute. (D.I. 130, 131, 134)
11

quantification of how the additional (or faster) mixing might affect the dissolution of
MFA, or the nucleation and crystal growth of MFM in Amneal's ANDA product. The
experts agree that the thermodynamic study does not inform this determination, as it is
not representative of the ANDA product (because of the addition of MFM) and did not
measure the effect of mixing speed or time on the rate of conversion. The expert
testimony - that conversion is system-dependent and the additional mixing performed
on Batch 16001 likely would have promoted conversion - renders any conclusion
regarding Batch 16001 Day 4 and Batch 16001A theoretical. 14 On the evidence
presented, the court concludes that Merck has not demonstrated that the additional
samples would yield different results. 15 Consequently, the court denies Merck's

14

The MOiier article (acknowledging funding by Merck) describes the selection and
SCXRD testing of a crystal to determine which polymorph was present in a sample of
Apotex nasal spray marketed in Canada. The testing identified an MFM crystal present
in the sample before the expiration date of the product. Peter MOiier, Mometasone
fuorate revisited, or how did the hydrate get in the bottle?, C71 Acta Cryst., 1080
(2015). (PTX 12) Merck argues that this article establishes "that MFA can convert to
MFM within the shelf life of an aqueous pharmaceutical suspension that is formulated
with MFA." (D.I. 163 at 16) Merck points to Amneal's counsel argument that the
formulation at bar is virtually identical to that found non-infringing in the Apotex litigation.
(D.I. 176 at 39:12-20); Schering Corp. v. Apotex Inc., 2012 WL 2263292 (D.N.J. June
15, 2012) ("the Apotex litigation"). The court does not find such argument persuasive as
the experts at bar have testified that conversion from MFA to MFM is particular to a
system. The court will not (without expert testimony) conclude that the formulation at
bar would behave in the same fashion as that tested in the article.
15

Nor has Merck demonstrated that Amneal's failure to produce additional samples of
Batch 16001 warrants an adverse inference. Indeed, Merck did not respond to
Amneal's arguments against such an inference in its reply brief. (D.I. 163 at 53-54; D.I.
165 at 54; D.I. 168 at 28-29); Braintree Labs., Inc. v. Novel Labs., Inc., 2015 WL
3492936 at *13 (D.N.J. June 1, 2015) (citation omitted) (An imposition of adverse
inference was not required when defendant "ha[d] not shown any prejudice, surprise, or
bad faith on the part of' plaintiff.).
12

alternative request for the production of 16001 Day 4 and 16001A samples and a new
trial. 16

4. Raman spectroscopy
Raman spectroscopy is a vibrational spectroscopy technique, which looks at the
way a molecule vibrates in a crystal. A microscope is coupled to an excitation laser and
a spectrometer. The laser is used to generate a Raman spectrum, which indicates the
vibrational modes of molecules and is used to differentiate between crystalline forms. A
Raman spectrum is a plot of the intensity (vertical axis) as a function of the vibrational
frequency (horizontal axis). Raman maps are a collection of Raman spectra (from a
hundred to over a million), which may then be analyzed by algorithms (or viewed
individually) to determine potential matches to a reference standard. Reference
standards are Raman spectra of known compositions. (D.I. 176 at 64-67, 78:20-79:6)

5. Merck's testing of the Exhibit Batches
Dr. Matzger used a Renishaw Raman spectrometer with point focus mapping
("point focus spectrometer"). This technique focuses the laser on a point and then steps
the laser to different positions on the sample to generate the Raman spectra. In
December 2015, Dr. Matzger switched to using a Renishaw Streamline® Raman
spectrometer ("Streamline spectrometer"), which allows for the collection of more data
by elongating the laser into a line and moving the sample continuously. (D. I. 176 at
75:20-23, 76:4-9, 106:1-4; see also D.I. 177 at 318:23-319:9) According to Dr. Matzger,

16

Merck asserts that whether Amneal should produce the samples is governed by
Federal Rule of Civil Procedure 26(b)(1 ), and contends that "it has established that the
additional mixing that the 16001 Day 4 and 16001A samples underwent increases the
likelihood of formation of MFM, thereby making those samples relevant to infringement."
(D.I. 168 at 26)
13

the Streamline spectrometer "spreads [the laser] power out more evenly over the
sample, so it tends to reduce the chance of damaging the sample with the laser." (D.I.
176 at 76: 18-20)
Dr. Matzger tested samples from four bottles of the Exhibit Batches over a period
of eight months starting in June 2015. He agreed that he did not find MFM in any of the
samples, and that some of the tested samples were over two years old. (Id. at 71: 1172:5, 87:5-9, 127:13-128:13; PTX 40) He testified that he used both point focus
spectroscopy (for samples tested prior to December 2015) and Streamline
spectroscopy (for samples tested after December 2015) to test the Exhibit Batches and
there was no difference in the results. (Id. at 123:20-24, 178:13-181:10)
The parties disagree on whether the Exhibit Batches are representative of the
ANDA product. 17 Merck argues that the Exhibit Batches are not representative of the
ANDA product, as they were not packaged according to the current Master Packaging
Batch Record and were mixed significantly less vigorously than the Batch 16001 Day 1.
(D.I. 163 at 18-19; D.I. 176 at 202-203) Dr. Matzger tested the Exhibit Batches, but did
not rely on the results for his ultimate infringement opinion. (D.I. 163 at 28) Amneal, on
the other hand, maintains that the Exhibit Batches are its ANDA product, arguing that it
used the Exhibit Batches to conduct testing (including the bioequivalence studies)
required by the FDA. (JTX 4, 5) For these reasons, Amneal references Dr. Matzger's
testing of the Exhibit Batches in its non-infringement arguments. (D.I. 165 at 31)

17

Merck states that on the record at bar, the Exhibit Batches were mixed at one-third to
one-half of the lowest acceptable values specified by Amneal's current ANDA. (D.I. 168
at 1-2) Although Amneal's counsel represented that typographical errors existed
regarding the mixing speeds, on the evidence presented, the court accepts Merck's
representations. (D.I. 165 at 15, 32-33)
14

There can be no dispute that Amneal provided data to the FDA obtained by
testing the Exhibit Batches. The court addressed the issue of additional (or faster)
mixing above. The court concludes that the testing of the Exhibit Batches (some of
which were tested after the two-year expiration) is relevant to the question of
infringement.

6. Merck's testing of Batch 16001 Day 1
Dr. Matzger tested seven slides prepared from one bottle of Batch 16001 Day 1.
Specifically, he briefly shook the glass bottle, used a pipette (or poured) to transfer
some material to the slide, allowed the slides to dry, and stored them in Petri dishes.
(D.I. 176 at 73:10-74:18, 101 :11-25) He performed Streamline spectroscopy on the
slides and generated Raman maps made up of millions of individual spectra. He
compared certain individual spectra (under 10% of spectra generated) to reference
spectra of MFM and MFA generated on the same equipment. 18 (Id. at 74:24-76:9,
78:20-79:22, 168:25-169: 16; PTX 40) He explained that MFM has a characteristic peak
(not shared) around 1709 cm- 1, MFA has a characteristic peak at 1725 cm-1, and the
two polymorphs share a peak in the range of 1640-1680 cm- 1 . (D.I. 176 at 81:17-83:4;
PTX 70)
By the characteristic peak at 1709 cm- 1, Dr. Matzger identified five M FM crystals
on three slides (the slide prepared February 8, 2015 had three particles and the slides

18

Dr. Matzger explained that a reference standard is collected with a high signal to
noise ratio, allowing the visualization of "all of the peaks that would be different between
the two forms" and the identification of "other maybe minor characteristic peaks." (D.I.
176 at 188:7-189:3)
15

prepared on March 13 and 19, 2016 each had one particle). 19 Each of the spectra
indicated the presence of both MFM (characteristic peak at 1709 cm- 1) and MFA (from
characteristic peaks including the one at 1725 cm- 1). 20 (D.I. 176 at 89-98, 158:4-13;
PTX 75-76, 79-80, 84-87) Dr. Matzger testified that he could not see any other
characteristic peaks for MFM "because of the signal to noise limitation." (D.I. 176 at
158:12-13, 160:22-161:5) He opined that noise occurs in all spectra and, in a
pharmaceutical product with a small amount of active ingredient, "noise will always be
present." (Id. at 190: 10-16) He explained that he took into account the signal to noise
ratio in his analysis and that a peak is considered significant if the ratio exceeds 2/1 or
3/1. 21 The signal to noise ratio in one spectra was 4.3/1 and, after "smoothing," it
"improved somewhat to" 4.7/1. (Id. at 93:6-24) Dr. Matzger explained that "smoothing"
increases the signal to noise value of a particular spectrum, at the expense of
resolution. It is accomplished by taking "three adjacent peaks and averag[ing] them

19

Dr. Matzger opined that he did not find MFM particles on the other four slides, but did
not perform extensive searches of those slides, as the searches were time consuming.
He "didn't see the point of doing additional searching" after identifying MFM on the other
slides. (D.I. 176 at 87-88, 153:24-154:9) He also explained that he would not have had
time to perform an extensive analysis of the other four slides. (Id. at 186:24-187:20) He
testified that when particles are very small (as in the samples at issue), "[t]he search
algorithms are just not that reliable," and potential matches must be inspected by hand.
(Id. at 67:25-68:5)
20

Dr. Matzger opined that none of the excipients would have a peak at 1709 cm- 1 . He
did not analyze pure samples of the excipients in the ANDA formulation, nor a placebo
containing all of the ingredients except the API to determine if any other peaks existed
in the general vicinity of 1709 cm- 1 to 1711 cm- 1 . He stated that he "had a number of
experiments where there was excipient and [he] did not identify peaks at that location."
(Id. at 165-167) He also relied on Amneal's ANDA stating that "excipient spectra will not
have peaks at wave numbers 1725±3 (or 1705), 1611 cm-1." (PTX 26 at AMLMF 4670)

21

Validation of Compendia! Procedures, USP 37. (PTX 67 at 3)
16

together." 22 He indicated for a particular spectra that "the[] two peaks are well resolved.
You can see the separation between them well enough that the reduction in resolution
is not an issue." (Id. at 90:6-91 :11; PTX 80) Dr. Matzger testified that he also
considered the width and shape of the peak to evaluate whether the peak was MFM as
opposed to an artifact. 23 (Id. at 94:7-20)
Amneal's expert, Dr. Brian Marquardt ("Dr. Marquardt"), analyzed the five spectra
from Dr. Matzger's report and opined that a "shoulder peak" indicative of MFA "could be
easily misinterpreted ... as a peak in that space and be misrepresented as MFM,"
given the concentrations and the signal to noise levels. He further explained that the
spectra show that "this is primarily the MFA form, which is indicated by th[e] secondary
doublet and the primary peaks ... , which are indicative of both forms. And it is my
opinion that Dr. Matzger misinterpreted this data as MFM." He concluded that MFM
was not present. (Id. at 310-311; DTX 138) Dr. Marquardt testified that in his study of
the API particles, the Raman spectra were "well above" a signal to noise ratio of 20/1.

(Id. at 308:5-10)
Dr. Matzger opined that a single characteristic peak is sufficient to distinguish
between MFM and MFA, as "based on the inspection of the reference patterns, you can
find one where they're well separated." When asked if a single peak was sufficient to
distinguish between MFM and MFA in Amneal's product, Dr. Matzger responded that
Amneal "did a similar analysis ... and ... also relied on a single peak in order to

22

In contrast, Dr. Marquardt opined that applying a smoothing algorithm to high
resolution Raman spectra, like that of Dr. Matzger, "can actually cause [one] to
misinterpret, possibly even generate [an] artificial peak." (D.I. 177 at 309:8-20)

23

Dr. Matzger opined that his spectra are "essentially the same" as certain of Amneal's
spectra from a limit of detection study. (D.I. 176 at 100:22-101:10; PTX 39, 84)
17

differentiate the monohydrate and the anhydrate." 24 He testified that he uses a single
peak "commonly to generate Raman maps." (D.I. 176 at 82-83; PTX 70) In contrast,
Dr. Marquardt testified that he "standardly use[s] three peaks" to identify a compound in
a complex mixture, because "using three peaks gives us an absolute confirmation of its
presence, and it provides us the specificity we need to confirm." He opined that it would
be appropriate to rely on one peak "at times," for example "in the determination of the
limit of detection ... where we add the material to the system, and we watch it actually
come in and grow into the system because we're adding known amounts." (D.I. 177 at
301:20-302:12)
Dr. Matzger testified that the particles of MFM were in the range of 1-2 microns,
but he was unable to measure the size. (Id. at 169:17-170:12) Dr. Marquardt explained
that using a scale factor, Raman instrumentation allows the visualization of particles that
are one to ten microns in diameter. (D.I. 177 at 323:7-16; see also 463:22-464:6)
Dr. Matzger testified that he would expect the amount of MFM to increase over
time due to crystal growth, but he did not analyze whether the identified MFM particles
did so. 25 (D.I. 176 at 129:11-14, 167-169) Dr. Rogers reviewed Dr. Matzger's
laboratory notebook, noting that Dr. Matzger prepared a slide on February 8, 2016 and
performed spectroscopy on it gathering data from February 8-10, and identifying three
micron sized particles of MFM. He testified that Dr. Matzger prepared slides on

24

Amneal performed and provided the results of certain testing to the FDA, including
Raman spectroscopy. (PTX 38) One of Amneal's scientist agreed that they "would look
at 1705 for monohydrate and 1725 for anhydrate" during testing. (D.I. 176 at 278:20279:4, 283:6-8; PTX 26, 38)

25

Merck points out that it asked Amneal multiple times if it would be allowed to
supplement the testing and Amneal refused, therefore, Merck did not request
supplementation of the expert record from the court. (D. I. 176 at 171 :3-23)
18

February 15, March 5, and March 12, and performed spectroscopy on each without
identifying any MFM particles. Dr. Matzger then prepared a sample on March 13 and
another on March 19. For each of these slides, Dr. Matzger performed spectroscopy six
days later identifying one micron-sized particle on each slide. Dr. Matzger prepared a
seventh slide on March 26, 2016 and performed spectroscopy on it, but did not identify
any MFM particles. Dr. Rogers disagrees with Dr. Matzger's conclusion about the
presence of MFM in Batch 16001 Day 1, explaining that "it can't be correct because if
MFM had been in there, it would have grown in both size and number. It would have
been easier to detect and he would have found a lot of it by the end." (D.I. 177 at
431 :9-433:5)
Dr. Matzger relied on his identification of one peak on a Raman spectra to
conclude that MFM particles were present in the tested Batch 16001 Day 1 samples.
He opined that a single peak is sufficient to identify MFM in the ANDA product, but
admitted that he could not see any other characteristic peaks for MFM because of the
signal to noise limitation. In contrast, Dr. Marquardt testified that three peaks are
generally used to identify a polymorph in an unknown sample. In Schering Corp. v.

Apotex Inc., 2012 WL 2263292 (D.N.J. June 15, 2012), the court evaluated expert
testimony regarding Raman spectroscopy results performed on the product at issue in
that case. The court concluded (based on expert testimony) that at least three peaks on

19

a spectra must be used to identify material based on accepted practices. 26 · 27 Id. at *710. The court assigns little weight to Dr. Matzger's identification of MFM based on a
single peak under these circumstances.

7. Amneal's testing of Batch 16001 Day 1
Dr. Marquardt28 analyzed ten samples from two bottles of Batch 16001 Day 1.
Specifically, five drops of material were placed on a slide, the slides were dried and
viewed under a HORIBA Scientific LabSpec 6 Raman microscope. Thirty API particles
were measured on two slides. Point focus spectroscopy was performed - two slides
with wide area mapping (4.5 micrometers steps) and six slides with high spatial
resolution mapping (2.5 micrometers steps). (D.I. 177 at 314-318; DTX 79, 133) Dr.
Marquardt analyzed the Raman spectra and compared them to reference spectra of
MFM and MFA, obtained using "individual points at the same accumulation time and
integration time." He examined 200 API particles using the three methods and did not
identify any MFM in the samples. (D.I. 177 at 323-326, 330, 333:3-334:7, 390:20-23)

26

Merck argues that the need for three peaks only applies to the x-ray crystallographic
powder diffraction pattern analysis opined on by the expert in the Apotex litigation and
not to Raman spectroscopy. The Federal Circuit heard the same argument from Merck
and subsequently affirmed the district court's judgment. See Merck Sharp & Dahme
Corp. v. Apotex Inc., 517 F. App'x 939 (Fed. Cir. 2013) (Rule 36 affirmance).
27

The court declines to reach Amneal's collateral estoppel argument.

28

Dr. Marquardt is the CEO and founder of MarqMetrics, a company providing
measurement solution services and developing Raman technology. (D.I. 177 at 297:17298:3) Dr. Marquardt's assistant at MarqMetrics, Dr. Andrew Knight ("Dr. Knight"),
prepared the slides and performed the Raman testing under Dr. Marquardt's direction.
Dr. Knight has experience in crystalline materials. (Id. at 327, 389:5-8) Having
reviewed Dr. Marquardt's background information, the court disagrees with Merck's
representation that Dr. Marquardt is only an expert in Raman spectroscopy "in a general
sense," and that neither he nor MarqMetrics has the requisite experience or skill to
conduct any meaningful tests of Amneal's ANDA product. (D.I. 163 at 38-41) The court
separately addresses Merck's specific concerns regarding Dr. Marquardt's testing.
20

The particles were selected by scanning the microscope stage in real-time and viewing
a "very noisy, very fast Raman spectrum" until observing "large indicator peaks" at 1470
cm- 1 and 1660 cm- 1 . (Id. at 326:8-327: 1) Dr. Marquardt testified that he was not "trying
to perform a massive amount of scans and not look at the data." Instead, he tried to do
"a fundamental objective design of experiments, where we could assess the quality of
the information that we collected, and then be able to make a conclusive determination
about whether there was or was not the presence of MFM in our Raman data." (Id. at
319:18-25) Dr. Marquardt agreed that his testing of Batch 16001 Day 1 reflects the
content of the sample at the time of testing. (Id. at 349:16-25)
According to Dr. Matzger, Dr. Marquardt did not "collect[] a sufficient number of
spectra [or] data ... on a significant number of particles within those spectra to draw the
conclusions that he has drawn." Moreover, "[i]f you want to say that something is not
present, it's pretty typical to do a limit of detection." (Id. at 483:18-485:4) Dr.
Marquardt, on the other hand, explained that "collecting millions and millions of spectra
doesn't do you much good if you don't actually analyze the data," and "taking a lot of
data doesn't say much about your scientific data. It just says you have an instrument
that can take data very quickly." (Id. at 331 :17-332:5) Dr. Marquardt agreed that the
detection limit of a technique determines whether a particle may be detected and that
he did not do a specific study on the effect of particle size on the detection limit.
However, he explained that none of the experts performed limit of detection studies.
(Id. at 395:22-397: 12)

Dr. Matzger testified that samples may be damaged by the laser during Raman
spectroscopy because the power densities are so high. A damaged sample cannot be

21

used for analysis and the damage is not always visible. He opined that a certain crystal
examined by Dr. Marquardt was visibly darker after testing, indicating that it was
burned. He testified that MFM is more thermally sensitive than MFA, such that less
power is needed to burn MFM. (Id. at 476-480) Dr. Marquardt admitted that he did not
specifically test the sensitivity of MFM to laser exposure. (Id. at 389:21-390:8) He
disagreed, however, with Merck's theory that his method "burned samples," explaining
that the power density of his point focus laser was in line with that used by Dr. Matzger
and his method was proper. Moreover, in preparation of a Raman method, they "always
perform a ... sample method prep where [they] look at laser power, ... at exposure
time, and ... actually test on a test sample to see if [they] are ... creating any damage
and ... to maximize our signal while minimizing the potential for damage in [the]
system." (Id. at 317:3-19, 340-341, 402: 1-403:3) He opined that the difference in color
of a certain sample was "a difference in contrast after ten-and-a-half hours of imaging
and bringing that image back in the microscope field of view." (Id. at 336: 14-19) On
cross-examination, he maintained that the difference in color does not "prove[] anything
besides the fact that [the pictures were] taken ten hours apart. The Raman spectra
were fine." (Id. at 379: 11-380:20)
Dr. Matzger also criticized Dr. Marquardt's use of a spike filter, opining that it
throws away data. For that reason, Dr. Matzger does not use a spike filter during data
collection for maps, but might use it during post processing when he is viewing the
spectra. (Id. at 480:5-21) Dr. Marquardt opined that a spike filter is commonly used in
Raman spectroscopy and allows the averaging out of random events. (Id. at 343:6344:4, 345:5-22)

22

Merck urges the court to assign no weight to Amneal's testing and related
testimony based on the fact that such experiments only indicate the presence or lack of
MFM at that specific point in time, not whether "MFM forms ... during the product's
proposed two-year shelf life." (D.I. 163 at 37) Neither expert has opined on a "growth
rate." In essence, two experts have tested the same samples and come (not
surprisingly) to opposite conclusions. Dr. Marquardt's testing is at least as relevant as
the conclusions drawn by Dr. Matzger.

8. Conclusion
Merck offers Dr. Matzger's testing of Batch 16001 Day 1 as persuasive evidence
of MFM in Amneal's ANDA product. Merck criticizes Dr. Marquardt's testing, arguing
that such testing is insufficient to prove a negative - that MFM is not present in the
ANDA product. The court observes that Dr. Matzger's testimony and Merck's related
arguments are self-serving, i.e., essentially arguing that Dr. Matzger's testing is "more
and better," therefore, only his opinion should be believed. The court is not so
convinced, and finds Dr. Marquardt's testimony at least as consistent and credible.
Weighing the evidence at bar (lack of MFM in the Exhibit Batches 29 and opposing
conclusions on the same testing 30 of the 16001 Batch Day 1), the court concludes that
Merck has not carried its burden to prove, by a preponderance of the evidence, that
MFM is present in Amneal's ANDA product during its two-year shelf life.

29

The court addressed Merck's concerns regarding the relevance of the Exhibit Batches
above.

30

The court recognizes that Dr. Matzger used a different testing protocol (Streamline
spectroscopy) and has addressed Merck's concerns about Dr. Marquardt's testing
above.

23

Ill. CONCLUSION
For the foregoing reasons, the court finds that Amneal does not infringe the
asserted claims. 31 An appropriate order shall issue.

31

The court does not reach Merck's contributory infringement arguments. (D.I. 163 at
58)

24