38 Hereditary Nonpolyposis Colon Cancer

(1)

38

Hereditary Nonpolyposis Colon Cancer

Lawrence C. Rusin and Susan Galandiuk

525

Colorectal cancer affects 148,300 patients in the United States

annually (72,600 males and 75,700 females) causing 56,600 deaths each year.

¹

Those patients who have two or more first- and/or second-degree relatives with colorectal cancer have a potentially definable inheritable disorder. Approximately 5%–6% of colorectal cancers have a known germline genetic mutation. Hereditary nonpolyposis colon cancer (HPNCC) is one of the syndromes and accounts for 3% of newly diag- nosed colorectal cancer cases.

²

HNPCC is characterized by the early onset of colorectal cancer (mean age, 45 years), with multiple generations affected. These cancers tend to be proximal to the splenic flexure, poorly differentiated, and have an increased fre- quency of synchronous and metachronous cancers. There is also an excess of extracolonic cancers, including endometrial, ovarian, gastric, small bowel, hepatobiliary, and transitional cell carcinomas. Over a patient’s lifetime, there is an 80% risk of cancer, with colon cancer being the most frequently diag- nosed cancer.

³

The syndrome is characterized by an autoso- mal dominant mode of inheritance. Germline mutations in mismatch repair (MMR) genes, which normally repair mistakes in DNA replication, are responsible for HNPCC.

Historical Perspectives

Alfred Warthin, a pathologist at the University of Michigan, first described a family with inherited intestinal cancer in 1895. Detailed descriptions of this family (known as “Family G”) as well as other families were published in 1913. Gastric cancer was the predominant cancer.

⁴

Nearly 50 years later, after spending time at the University of Michigan, Henry Lynch published the pedigrees of two large midwestern fam- ilies (“Family N” from Nebraska and “Family M” from Michigan). In his article, he commented on the wide spectrum of cancers diagnosed and the probability of an inheritable gene. By the mid-1980s, after further studies, two patterns of disease presentation became apparent, so-called Lynch I (col- orectal cancer only) and Lynch II (colorectal and other malig- nancies). Concurrent observations determined that there was

a paucity of colonic polyps in these patients, and that there was considerable overlap between Lynch I and II syndromes.

The syndrome became known as HNPCC.

⁵

To promote research collaborative studies, the International Collaborative Group on HNPCC met in Amsterdam in 1990.

⁶

A set of diagnostic guidelines was agreed on that would allow researchers to gather homogenous populations to be studied.

Once HNPCC was well categorized, rapid progress was made elucidating the genetic defect as well as in diagnosis and treatment of the disease. Colorectal tumors were found to have multiple mismatched nucleotides. This unique genetic abnormality was termed replication error phenotype or RER+. Most of these mismatched bases were in areas of the gene called “microsatellites,” so the term microsatellite insta- bility (MSI) became prevalent.

⁷

Extensive research in yeast and Escherichia coli had identified a group of genes referred to as MMR genes. Using linkage studies, the first human homolog, MSH2, was identified in 1993 by Fishel.

⁸

Genetics

What Are Microsatellites?

Microsatellites are short tandem repeating base sequences that are usually mononucleotide or dinucleotide base repeats.

Most often, the repeats are found in the noncoding or intronic portion of the gene. However, microsatellites can occur any- where within the gene. There are well over 200 polymorphic microsatellite loci identified.

⁹

The most common sequences are repeats of adenine or thymine or the dinucleotide repeats of cytosine/adenine (CA

_n

) or guanine/thymine (GT

_n

). When the number of repeats in a microsatellite sequence in a cancer cell is different from the surrounding normal tissue, this is termed “microsatellite instability.”

Different Types of DNA Repair

There are several intracellular mechanisms that repair DNA

damage and maintain genomic stability and fidelity. “Base

excision repair” (BER) repairs mutations caused by reactive

(2)

oxygen species related to aerobic metabolism. Oxidative DNA damage produces a stable 8oxo 7-8 dihydro-2′

deoxyguanosine (8 oxoG), which readily misrepairs with ade- nine residues resulting in G:C→T:A transversion mutations.

“Nucleotide excision repair” (NER) repairs damage caused by exogenous agents such as mutagenic and carcinogenic chemicals as well as ultraviolet radiation. Several genes have been cloned XPA through XPG. MMR genes repair single base mismatches as well as insertion/deletion loops (IDL) of up to 10 nucleotides. MMR gene dysfunction is characterized by MSI and is responsible for HNPCC. The so-called mutator phenotype of HNPCC is characterized by an increased genome-wide mutation rate. This is in contradistinction to sporadic colon cancer or familial polyposis, which is charac- terized by loss of whole portions of chromosome alleles known as loss of heterozygosity (LOH).

¹⁰

MMR Function in Single Cells

This MMR repair system has been well studied and eluci- dated in single cell organisms. There are three main compo- nents of this repair system, namely, MutS, MutL, and MutH.

MutS recognizes base mismatches and IDL. Acting as an adenosine triphosphatase (ATPase) homodimer, MutS attaches to the abnormal DNA, undergoes a conformational change, and translocates or slides along the DNA, allowing the formation of a large DNA loop. As the DNA is being shortened into the loop, the abnormal mismatches are included in the DNA loop. MutL, another ATPase, again act- ing as a homodimer, identifies the loop segment of DNA that includes the mismatch. The loop continues to enlarge, and a DNA-methylated (DAM) GATC base sequence is encoun- tered. The new DNA strand is recognized because it is not methylated. At this point, MutH, an endonuclease, is activated and a nick in the DNA strand is made. This can occur on either the 3′ or 5′ side of the DNA strand. Bidirectional exonucleases excise the abnormal DNA strand and DNA polymerase resynthesizes a new strand.

¹¹

MMR Function in Humans

MMR genes have been found in every organism studied and are homologous with yeast and bacterial MMR genes. In humans, five MutS genes have been identified (MSH2, MSH3,

MSH4, MSH5, MSH6) and four MutL genes (MLH1, PMS1, PMS2, MLH3). No MutH gene equivalent has been discov-

ered. In humans, MutS and MutL presumably directly activate exonucleases without the need for a MutH gene.

Unlike single cell MMR genes that function as a homod- imer, human MMR genes function as a heteroduplex (Figure

38-1). MSH2 acts as a “scout” and identifies the mismatches

in the new DNA strand. It then complexes with MSH6 to form the MutSα complex that identifies single nucleotide misre- pairs. Alternatively, MSH2 can complex with hMSH3 form- ing MutSβ complex that repairs IDL with up to 10 base pairs.

These pathways are not mutually exclusive as MutSα can rec- ognize 2–4 base IDL, but there is less affinity. Likewise, MutSβ can recognize single base pair insertions and dele- tions. Both MSH3 and MSH6 must be abnormal to have complete loss of hMSH2-dependent MMR.

¹²

MLH1 and PMS2 bind to form a second heteroduplex that interacts with the MutS duplex, stimulating excision and resynthesis. The exact mechanism of action of the human MutL MMR genes is not known. The most frequently mutated genes in HNPCC are MLH1 (33%) and MLH2 (31%) and rarely PMS2 or PMS1.

¹³

Of the mutations identified, 90%

occurred in MLH1 and MSH2.

¹⁴

Recently, a metaanalysis of index families fulfilling the Amsterdam criteria, worldwide, revealed that a mutation in MLH1 is found in 25.5%–29.6% of families, and MSH2 is found in 14.8%–21.6% of families. In general, the overall mutation discovery rate is less than 50%.

¹⁴

MutS has two functions—mismatch recognition and ATPase activity. Both functions are essential for repair in 7 of 8 possible single-base mismatches. The C:C misrepair is not recognized, but fortunately occurs least frequently.

¹⁰

It is gen- erally agreed that mismatched DNA stimulates ATP hydroly- sis as part of the repair mechanism, but there is disagreement as to the role this has in MMR gene function. It is agreed that

CAAGTCC

CAAGTCC CAAGTCC

GTTCAGG

GTTCAGG GTTCAGG

GTTCAGG

GTTCAGG 1-bp mismatch

3-bp mismatch

CAATTCC

hMutSα

hMutLα hMutSβ

hMSH2

hMSH2 hMSH2 hMSH6

hMLH1

hMLH1 hPMS1

hPMS1

hMSH3

hMSH3 hMSH6

TAG

FIGURE38-1. The DNA MMR system can correct either single base- pair mismatches or larger loops of mismatched DNA. hMSH2 serves as the “scout” that recognizes mismatched DNA. It forms a complex with either hMSH6 or hMSH3, depending on the number of mismatched nucleotides. A second heterodimeric complex (hMLH1/

hPMSI) is then recruited to excise the mispaired nucleotides.

hMUTSα =hMSH2/hMSH6; hMuTSβ =hMSH2/hMSH3; hMutLα

=hMLH1/hPMS1. bp =base pair. (Reprinted with permission from Chung and Rustgi.²⁹)

(3)

the MutS with adenosine 5′-diphosphate (ADP) has the highest affinity for the mismatch.

In the translocation model, ATP enables MutSα or β to move like a motor protein away from the mismatch. This allows formation of an α loop similar to that described in the

E. coli model. When the MutS ADP is attracted to the

MMR, it is converted to MutS ATP, which has reduced affin- ity for the mismatch and activates the binding sites for translocation along the newly synthesized DNA. This draws flanking DNA toward the mismatch, causing it to form into an α loop.

¹⁵

The molecular switch theory proposes that this transposi- tion from ADP to ATP in the MutS gene switches the repair complex on and off. MutS ADP has a high affinity for the misrepaired gene and then it binds to it. When this occurs, ADP is transformed to ATP and causes a conformational change in the MutS complex. A clamp is formed around the mismatched DNA. Several of these clamps surround the mis- repair and signal either a repair process or induce apoptosis via a P53 independent pathway. When ATP is hydrolyzed to ADP, the clamp is switched off and is released from the DNA strand. The MutS complex is now recycled and can recognize a new mismatch. The ATP hydrolysis in this model turns the MMR switch off and on but is not intrinsically involved in the actual repair.

¹⁶

The third theory is called “transactivation.” MutS binds both ATP and the mismatch DNA simultaneously in order to activate MMR. The ATPase in this model functions as a proofreader and binds only to heteroduplex DNA. Once the mismatch and the ATP are bound, it is believed that a signal is produced to authorize DNA repair.

¹⁷

Genes Implicated in HNPCC Carcinogenesis

When MMR genes are dysfunctional, the myriad genes that have microsatellites in their coding region can become inacti- vated, causing acceleration of tumorigenesis. Transforming growth factor beta Type II receptor gene (TGFβRII) has a 10 adenine-repeating tract in its coding area. TGFβRII was found to be mutated in 70%–90% of cancers with MSI.

^18,19 TGFβRII acts as a tumor suppressor gene and is a strong

inhibitor of epithelial growth. Insulin-like growth factor II receptor gene (IGFRII) binds insulin-like growth factor I, thus antagonizing its growth stimulating effects. IGFRII also acti- vates TGFβI from its latent to active form. There is a long 8 repeat guanine tract in IGFRII. In 90% of tumors studied, either

TGFRII or IGFRII was abnormal but not both. This suggests

that the genes are part of the same tumorigenesis pathway.

²⁰

The BAX gene is another gene found to be mutated in up to 54% of MSI-H colorectal tumors.

¹⁹

It is involved in apoptosis and is a downstream gene often activated by the P53 gene. It is of note that MSI-H tumors have wild-type P53. In the cod- ing region of the BAX gene, there is an 8 guanine repeating tract spanning codons 38–41 that is highly susceptible to frameshift mutations.

²¹

Other genes with microsatellites in

the coding region are MSH3, MSH6, TCF4, BLM (19),

Capase-5,²²

and the PTEN gene and APC.

²³

Because only 50% of HNPCC tumors have mutations in known MMR genes, a search for other unique gene pathways that could affect DNA repair is ongoing. RAF oncogenes are a family of genes that encode kinases and mediate cellular responses to growth signals. BRAF is a RAF gene that is com- mon in MMR-deficient tumors and in colorectal tumors that have a normal KRAS gene. Both BRAF and KRAS mutations are found in all stages of colon cancer and in adenomas greater than 1 cm.

²⁴

Another DNA repair gene is MED1 that has endonuclease activity and could function as a MutH gene.

It has four hypermutable tracts of adenine repeats. MED1 alteration may contribute to progressive MSI as part of

“mutators mutation” phenomena.

²⁵

Pathologic Features

There are unique pathologic features associated with the mutator phenotype of HNPCC. Cancers are often mucinous or poorly differentiated with signet ring cells.

²⁶

Despite what appears to be unfavorable histology, the incidence of lymph nodes is 35% compared with 65% in sporadic colon cancer.

²⁷

In 9% of patients, there is an extensive peritumoral lympho- cytic infiltration that may be the result of an enhanced immunologic response (Figure 38-2A). Others have described a Crohn’s-like lymphocytic reaction with the accumulation of both B and T lymphocytes around the tumor (Figure 38-2B).

²⁸

Flow cytometry showed most HNPCC tumors to be diploid.

Because HNPCC cancers have small unstable tracts, large chromosomes are not lost. This is in contradistinction to the frequent aneuploidy of sporadic colon cancer where tumori- genesis is related to the LOH.

²⁹

Clinical Features of HNPCC

Affected individuals in HNPCC have an increased risk of colon cancer and other extracolonic cancers as demonstrated in Table 38-1. Colon cancer is the most frequently diagnosed cancer in HNPCC (80%) and endometrial cancer is the most frequent extracolonic cancer (50%–60%).

²⁹

Colorectal can- cers in HNPCC are proximal to the splenic flexure (68% in HNPCC versus 49% of sporadic cancers), are more likely to have associated synchronous cancers (7% HNPCC versus 1%

sporadic colon cancer), and will have increased metachronous cancers at 10 years (29% HNPCC versus 5% sporadic can- cers).

³⁰

Similarly, women with HNPCC-related endometrial cancer have a 75% risk of a second cancer during a 26-year follow-up. The median age of onset of colon cancer is 42 years and for endometrial cancer it is 49 years.

³¹

Despite being termed hereditary nonpolyposis colon cancer,

a polyp is still the precursor lesion for a colorectal cancer. The

phenotype is not the extensive polyposis as seen in familial

(4)

adenomatous polyposis (FAP). In a Danish surveillance study, 70% of mutation carriers developed adenoma by age 60 as opposed to 37% in the control group. The adenomas in the mutation carrier group were larger and had a higher propor- tion of villous components and high-grade dysplasia.

Adenomas were located in the proximal colon and 70% of the polyps had absent MMR genes on immunohistochemistry.

³²

One cancer is prevented for every 2.8 polyps removed in HNPCC patients

³³

compared with one cancer being removed for every 41–119 polypectomies in the general population.

³⁴

It is estimated that malignant transformation occurs in 3 years in HNPCC as opposed to 10 years in sporadic colon cancer.

In other studies, HNPCC polyps are found throughout the colon and not necessarily clustered near the proximal colon where the predominance of cancers are seen.

³⁵

Two other types of polyps—the flat adenoma and serrated adenoma—have been implicated as possible precursors of the HNPCC cancers. Flat adenomas are found proximally in up to 50% of HNPCC patients

³⁶

(Figure 38-3A and B).

Approximately 20% of flat adenomas will be MSI-H and will also have a mutation in the TGFβRII gene.

³⁶

These polyps are difficult to detect during colonoscopy without dye spray tech- niques (e.g., methylene blue). Furthermore, flat adenomas with advanced histology (high-grade dysplasia or cancer) are significantly smaller (10.75 mm) than comparable polypoid lesions (20 mm).

³⁷

It has been proposed that hyperplastic polyps, serrated adenomas, and serrated adenocarcinomas form a morphologic continuum. Carcinomas associated with serrated adenomas have a predilection for the cecum and the rectum and are often mucinous in nature. MSI is found in 37% of the cases.

³⁸

Genotype–Phenotype Relationships

There are very few studies that evaluate phenotype–genotype correlations. One study of 35 families found the MSH2 muta- tion to be associated with a late age onset of rectal cancer and more extracolonic cancers than in the MLH1 mutation- positive group.

³⁹

Mutations in MSH2 were associated with an increased risk of rare extracolonic cancers in another study.

⁴⁰

Amsterdam-positive families with known mutations were compared with Amsterdam-positive families without a known mutation. The subgroup without mutations had an increased risk of rectal cancer, fewer HNPCC-related cancers, a lower frequency of multiple colorectal cancers, and a later age of onset.

⁴¹

The Muir-Torre syndrome is characterized by sebaceous adenomas, sebaceous carcinomas, and keratoacanthomas associated with multiple visceral tumors. Colorectal cancers are most frequently found (51%) and are often proximal to the splenic flexure (60%). Although only 25% of Muir-Torre patients develop a polyp, 90% of patients who develop polyps

FIGURE38-2. A Medullary carcinoma type pattern with peritumoral

lymphocytic infiltrate. B MSI-H cancer with marked peritumoral lymphocytic infiltrate (Crohn’s-like reaction), ×20 magnification.

(Courtesy of Robert E. Petras, MD, National Director Gastro- intestinal Pathology Services, Ameripath Inc., Oakwood Village, OH and Associate Professor of Pathology, Northeastern Ohio University College of Medicine).

TABLE38-1. Lifetime risks for cancer associated with the hereditary nonpolyposis colorectal cancer syndrome

Type of cancer Persons with HNPCC General population

Colorectal 80–82 5–6

Endometrial 50–60 2–3

Gastric 13 1

Ovarian 12 1–2

Small bowel 1–4 0.01

Bladder 4 1–3

Brain 4 0.6

Kidney, renal, pelvis 3 1

Biliary tract 2 0.6

Source: Reprinted with permission from Chung and Rustgi.²⁹

(5)

will develop colon cancer. The second most frequent tumors are genitourinary (24%).

⁴²

Germline mutations in MLH1

⁴³

and MSH2

³⁹

have been identified. As expected, many of the tumors exhibit MSI. The visceral tumors are often low grade, and prolonged survival in the presence of metastatic disease has been reported. The median age of diagnosis is 55 years and only 60% will have a positive family history.

⁴²

Aggressive screening and surveil- lance for colorectal and genitourinary cancers as recom- mended for HNPCC are warranted.

MUTS

α

(a heterodimer of MSH6 and MSH2) senses and repairs single base mismatches. In a study of 91 familial

non-HNPCC patients, six patients (7.0%) were found to have

MSH6 mutations but none were found in 58 HNPCC patients.

These tumors were microsatellite low (MSI-L), had a median onset of 61 years, and did not fulfill classic Amsterdam crite- ria.

⁴⁴

Other studies of 90 mutation-negative HNPCC families and HNPCC-like families found only one MSH6 mutation.

⁴⁵

Despite this discrepancy, the International Collaborative Group has collected more than 30 potentially pathogenic

MSH6 mutations. Thirty-five percent of these mutations

involve only one amino acid.

⁴⁶ MSH6 is the third most fre-

quently affected gene in HNPCC. Because MutSβ (het- erodimer of MSH2 and MLH1) can act redundantly to repair single base mismatches, and insertion–deletions of large din- ucleotide sequences is intact, there are less frame shifts. This results in less loss of function in affected genes. A weaker phenotype occurs with slower tumorigenesis and cancers that are MSI-L.

⁴⁵MSH6 knockout mice display an MSI-L pheno-

type and the majority of tumors in mice are endometrial or cutaneous with few intestinal cancers.

⁴⁷

Colorectal cancers are more frequently left-sided in MSH6 carriers.

⁴⁸

The risk of endometrial cancer is increased over

MSH2 or MLH1 carriers (76% versus 30%) and colon cancer

is decreased (32% versus 80%). The median age of onset is 55 years.

⁴⁹

Therefore, a distinct phenotype has emerged, char- acterized by a clustering of endometrial cancer, a decreased frequency of colorectal cancer, later age of onset, and MSI-L cancers. Some MSH6 cancers are MSI-H. Combined MSI test- ing and immunohistochemistry, as well as a distinct pheno- type, are used to select families for MSH6 mutation analysis.

⁴⁹

Diagnosis

The key to the diagnosis of HNPCC is a detailed family his- tory and subsequent construction of a family pedigree.

Amsterdam criteria (Table 38-2) were created to define clini- cal criteria to identify patients with a high probability of

FIGURE38-3. A Colonoscopic view of a flat adenoma in the cecum

that could easily be overlooked. Such polyps are more easily seen using dye-spraying techniques. B Microscopic view of same polyp after endoscopic removal, showing severe dysplasia, ×100 magnification. (Courtesy of Dr. Robert E. Petras, MD, National Director Gastrointestinal Pathology Services, Ameripath Inc., Oakwood Village, OH and Associate Professor of Pathology, Northeastern Ohio University College of Medicine).

TABLE38-2. Amsterdam criteria Amsterdam criteria

● At least three family members with colorectal cancer, one of whom is first-degree relative of the other two.

● At least two generations with colorectal cancer.

● At least one individual < 50 y at diagnosis of colorectal cancer.

Amsterdam criteria II

● At least three family members with HNPCC-related cancer, one of whom is first-degree relative of the other two.

● At least two generations with HNPCC-related cancer.

● At least one individual < 50 y at diagnosis of HNPCC-related cancer.

Modified Amsterdam criteria

● Two first-degree relatives with colorectal cancer involving two generations.

● At least one case diagnosed before 55 y or

● Two first-degree relatives with colorectal cancer and a third relative with endometrial cancer or another HNPCC-related cancer.

Source: Modified with permission from Chung and Rustgi.²⁹

(6)

having an inheritable form of colon cancer not associated with diffuse polyposis. For day-to-day clinical purposes, the original definition was too restrictive because it did not account for the extracolonic malignancies associated with HNPCC, the decrease in the average family size, the late onset variants of HNPCC, and the problems with incomplete data recovery. Modifications of the Amsterdam criteria have occurred and are known as Amsterdam II and modified Amsterdam criteria (Table 38-2). A National Cancer Institute workshop on HNPCC was held in 1996. Out of this meeting, a broader, less-specific set of guidelines was introduced (Table 38-3).

⁵⁰

These criteria, known as the Bethesda criteria, were formulated to increase sensitivity at the expense of specificity and focused on individual patients. Because MSI status is linked to HNPCC and the clinical criteria were broadened, MSI testing of this population can define a sub- population for genetic testing.

It is important to remember that HNPCC is a clinical diag- nosis and genetic testing cannot prove a family does not have HNPCC. Gene testing can potentially decrease the cost and morbidity of screening and surveillance, help with surgical decisions, stratify patients for chemoprevention, and help patients cope with job and family planning.

Table 38-4 compares the sensitivity and specificity of sev-

eral of the clinical criteria for identifying a pathologic muta- tion. This table is derived from a cohort of 70 families. Only 18 families had a mutation found and six of these were incon- clusive.

⁵¹

Genetic Testing

Gene sequencing of MSH2 and MLH1 is now commercially available. This testing may not be covered by insurance com- panies and the cost is approximately $1750. Furthermore, interpretation of gene testing is complicated because muta- tions are spread diffusely throughout MSH2 and MLH1.

Mutations themselves are a broad spectrum of truncating, frameshift, and missense mutations. A missense mutation results in the substitution of a single amino acid, and the effect on protein function may be negligible. This type of mis- sense mutation is often reported as a mutation of unknown significance. Approximately 31% of MLH1 mutations are this

type.

⁵²

Missense mutations must meet strict criteria to be con- sidered pathologic: Is the amino acid change conserved across species? What is the mutation population frequency? Does the mutation segregate with cancer in kindreds? Finally, does the alteration affect gene function? A data bank of known mutations is kept by the International Collaborative Group of HNPCC now known as InSiGHT.

Only half of well-characterized HNPCC families have a mutation identified. This suggests that there are other unknown MMR genes or that current techniques are lacking in sensitivity. Before genes are sequenced, they are amplified by polymerase chain reaction (PCR) and both alleles are sequenced simultaneously. A wild-type allele may mask a very short mutated allele. A technique called conversion has been used to detect these hidden mutations. Patient mononu- clear cells are fused with a specially designed rodent cell line.

One quarter of the hybrids contain a single copy of the human genes, thus the human gene is now haploid and can be ana- lyzed and characterized.

⁵³

In a series of 10 suspected HNPCC families, no mutations were found on initial gene sequencing, but a mutation was found in all 10 when the conversion technique was used.

⁵⁴

Once criteria to identify HNPCC families had been broad- ened, direct gene testing by sequencing was less effective (Table 38-4). Two techniques that are more cost effective—

MSI testing and immunohistochemistry—are performed before gene sequencing if the strict Amsterdam criteria or the Bethesda 1–3 criteria are not fulfilled. MSI testing is done on tissue. An international workshop was held and five validated microsatellite markers were selected to ensure uniformity of diagnosis between different laboratories. These microsatellite loci include two polyadenine sequences (BAT25 and BAT26) and three dinucleotide repeat sequences (D2S123, D5S346, D172S250).

⁵⁵

Microsatellite testing can be done on fresh tissue or paraffin blocks. A typical test for MSI is seen in

Figure 38-4. If none of the markers are unstable, the tumor is

called microsatellite stable (MSS). When only one marker is unstable, this is termed MSI low (MSI-L), and microsatel- lite high (MSI-H) will have two or more markers positive.

When a tumor is MSI-H, then gene sequencing of MSH2 or

MLH1 can be performed. The cost of MSI testing is $700.

Immunohistochemistry is a technique using antibodies to MSH2 and MLH1. This is a standard laboratory procedure (Figures 38-5 and 38-6) to identify which MMR gene is abnormal so that only one MMR gene needs to be sequenced.

TABLE38-3. Modified Bethesda guidelines Amsterdam criteria

●Patient with two HNPCC-related tumors.

●Patient with colorectal cancer with first-degree relative with HNPCC- related cancer; one of the cancers at < 50 y or adenoma at < 40 y.

●Patient with colorectal cancer or endometrial cancer at < 50 y.

●Patient with right-sided, undifferentiated colorectal cancer at < 50 y.

●Patient with signet ring colorectal cancer at < 50 y.

●Patient with adenoma at < 40 y.

Source: Modified from Rodriguez-Bigas et al.,⁵⁰ with permission from Oxford University Press.

TABLE38-4. Direct mutation finding (n =70)

Category Sensitivity (%) Specificity (%)

Amsterdam [n =28] 61 67

Amsterdam II [n =34] 78 61

Bethesda [n =56] 94 25

Bethesda (1–3) [n =44] 94 49

Source: Adapted and reproduced from Syngal et al.,⁵¹with permission from the BMJ Publishing Group.

(7)

If immunohistochemistry reveals that one gene is absent, this is presumptive evidence that the patient probably has HNPCC. Recently, 28 families with a known MSH2 or MLH1 mutation were studied. The MSI panel identified all families as being MSI-H, but in five cases a mutant protein was

expressed that could be detected by immunohistochemistry thereby producing a false test.

⁵⁶

A cost analysis has been performed on four different gene testing strategies: 1) gene sequencing for Amsterdam-positive individuals (“Amsterdam” strategy); 2) tumor MSI analysis of individuals who meet less stringent modified clinical criteria

FIGURE38-4. Detection of MSI with the use of fluorescent labeling of PCR products analyzed in an automatic sequencer. Two markers are analyzed in the same track: the mononucleotide repeat marker BAT26 is shown on the left, and the dinucleotide marker D2S123 is shown on the right. The upper tracking is from germline DNA from blood. The lower tracing is from DNA extracted from a histologic section of a tumor containing more than 50% tumor cells. For marker BAT26, germline DNA shows a single peak, indicating that the patient is homozygous for this marker (arrow). Tumor DNA shows, in addition to the normal allele (single arrow), a new allele (double arrows) that has lost approximately five nucleotides. This constitutes microsatellite stability. For marker D2S123, germline DNA is homozygous, whereas tumor DNA shows two new alleles (triple arrows), one with a loss of approximately 10 nucleotides (left) and one with a gain of two nucleotides (right). Thus, the tumor shows MSI with both markers. All peaks display “stutter”—that is, small amounts of material with a gain or a loss of one or a few nucleotides. This is a normal phenomenon. (Reprinted with permission from Lynch HT, De la Chapelle A. Hereditary colorectal cancer. N Engl J Med 2003;348:919–932. Copyright © 2003 Massachusetts Medicine Society. All rights reserved).

FIGURE38-5. hMLH1 immunohistochemistry. Blue arrow indicates positive nuclear staining for the presence of hMLH1 protein within an inflammatory cell. Black arrow demonstrates the absence of protein within cancer cells, ×400 magnification. (Courtesy of Robert E.

Petras, MD, National Director Gastrointestinal Pathology Services, Ameripath Inc., Oakwood Village, OH and Associate Professor of Pathology, Northeastern Ohio University College of Medicine).

FIGURE38-6. hMSH2 immunohistochemistry. Positive nuclear staining demonstrates the normal presence of hMSH2 protein, ×400 magnification. (Courtesy of Robert E. Petras, MD, National Director Gastrointestinal Pathology Services, Ameripath Inc., Oakwood Village, OH and Associate Professor of Pathology, Northeastern Ohio University College of Medicine).

(8)

and germline mutation of individuals with MSI-H tumors (“modified” strategy); 3) MSI testing of all colorectal cancers and germline analysis of individuals with MSI-H tumors (“test-all” strategy); and 4) germline mutation analysis of individuals who satisfy the Amsterdam criteria and tumor MSI analysis of the remaining individuals who meet less strin- gent criteria (“mixed” strategy). The least expensive strategy was the Amsterdam only strategy, but only 10% of mutations were identified. The modified and mixed strategies identified 60% and 75% of the mutations, respectively. These strategies were $400.00 less expensive than the test-all strategy in the model. Only seven mutations were missed by the mixed strat- egy per 1000 colorectal cancers. The cost for finding each of these seven extra mutations in the test-all strategy was $51,000 per mutation.

⁵⁷

Currently, a mixed strategy seems most appro- priate. MSI testing of all colorectal cancers may become stan- dard in the future because of its potential relationship to increased survival and response to chemotherapy. This would make testing all colorectal cancers the best medical practice.

Before genetic testing in HNPCC is done, informed consent must be obtained. Testing should begin with an affected indi- vidual (an HNPCC cancer has been diagnosed). It is best to choose a proband with the earliest age of onset of cancer. For a patient who meets the Amsterdam II criteria or the first three Bethesda criteria, it is reasonable to proceed to germline test- ing. When the clinical risk is lower, MSI testing or immuno- histochemistry should be performed first. If the tumor stains positive for MSH2 and MLH1 or is MSS or MSI-L, no further genetic testing is performed. However, if the tumor is MSI-H, or there is absent staining on immunohistochemistry for MLH1 or MSH2, then gene sequencing is performed. Note that only 50% of Amsterdam families have an MMR mutation on gene sequencing, and that HNPCC is a clinical diagnosis, not a genetic diagnosis. Further genetic testing can be performed in a research setting when mutations in MSH2 and MLH1 are not found. Conversion testing as previously described can be per- formed as well as testing for hMSH6 (especially in MSI-L patients or when there are late onset HNPCC pedigrees).

Because of the high number of missense mutations in HNPCC, the gene-sequencing test may be reported as a muta- tion of unknown variance. This is a non-informative test.

When the proband has a negative or a non-informative test, no further gene testing in the at-risk family members is possible.

All family members require intensive surveillance. When the proband has a positive mutation, the family members that test positive are carriers of the abnormal gene and need extensive surveillance and/or prophylactic surgery. However, when the proband’s test is positive and the at-risk family members are mutation negative, these family members are not carriers of an MMR gene. Their colon cancer risk is that of the general population. They should be screened accordingly. The cost of screening family members for a known mutation is $300.

There are also a number of ethical considerations regarding the ordering of genetic testing with respect to insurance

discrimination. Currently, nearly all states have passed legis- lation prohibiting genetic discrimination in health insurance;

however, only “asymptomatic” individuals are protected by such legislation.

^58,59

Registries

Once a diagnosis of HNPCC is suspected, a referral to a reg- istry for inherited colorectal cancer or to a cancer center with a high-risk clinic is important. A registry has a database or list of families and their members who have a high frequency of colorectal cancer. Personnel at the registry consist of a coordi- nator, oftentimes a genetic counselor, and a physician director.

The whole focus of the registry is patient care, education of patients, families, and referring physicians, and promotion of collaborative research. No single physician has the time to educate high-risk individuals about autosomal dominant inheritance and the cancer risk of inherited colorectal cancers;

to gather factual medical data about widely dispersed family members and construct a pedigree; to give counsel about the risks and benefits of gene testing; to perform gene testing and carefully and confidentially explain the results; and finally to coordinate treatment and life-long surveillance. Registry per- sonnel provide further support for myriad problems such as family stress, health insurance difficulties, and job discrimina- tion. A local registry can be found by accessing the Collabo- rative Group of the Americas on Inherited Colorectal Cancer at http://www.cgaicc.com. To emphasize the need for expert counseling and care, a study of 177 patients undergoing genetic testing for FAP was performed. Of those patients, only 18% received pretest counseling, 16% gave a written consent, and a full 30% were given incorrect test interpretation.

⁶⁰

Surveillance

In 1997, a number of medical societies proposed uniform

screening criteria for colorectal cancer, proposing separate

criteria for average- and high-risk patient groups. Patients

with HNPCC or a family history thereof constitute one of

these high-risk groups. Because of the 80% lifetime risk of

developing colorectal cancer, the American Cancer Society

recommends colonoscopy every 2 years beginning at age 21,

and annual colonoscopy recommended beginning at age 40.

⁶¹

Because cancers in HNPCC can occur at very young ages and

develop within 2 years of a negative colonoscopy,

⁶²

it is

important to stress to patients that if they have symptoms,

they should be checked regardless of their age and even if

they have had a recent negative colonoscopy. Colonic neopla-

sia in individuals with HNPCC seems to develop more rapidly

than in average-risk nonaffected individuals.

⁶³

For this reason,

many, including the authors, believe that colonoscopy in these

individuals should be performed annually rather than every

(9)

2 years, even in young individuals. Adenomas in HNPCC patients occur earlier and are more likely to be villous.

⁶⁴

The value of screening colonoscopy in HNPCC was demonstrated by Jarvinen and colleagues

⁶⁵

who studied a group of 252 individuals belonging to 22 HNPCC families.

Of these, 137 participated in screening colonoscopy every 3 years, whereas the remainder refused such evaluation.

Colorectal cancer developed in 8% of the screened family members, compared with 16% of those who refused screen- ing. In those individuals who were known to have a DNA MMR gene mutation, the rate of colorectal cancer in those who underwent screening was 18% compared with 41% in those who did not undergo screening. All cancers that devel- oped in the screened group were either Dukes’ A or B lesions, with no attributable deaths, compared with more advanced lesions in the unscreened group and nine deaths attributed to cancer (8%). Jarvinen and colleagues demonstrated that colonoscopic screening at 3 yearly intervals in HNPCC fam- ily members not only reduced the risk of colorectal cancer by half but prevented deaths caused by colorectal cancer and reduced the mortality by two-thirds. Similarly, the study by Rekonen-Sinislao and associates

⁶⁶

also demonstrated an ear- lier cancer stage at diagnosis and better 10-year survival in those individuals in whom colorectal cancer was detected dur- ing a colonoscopic screening examination compared with those in whom diagnosis was made based on symptoms.

Because of the high risk of endometrial cancer in women, annual transvaginal ultrasound to examine the endometrium, preferably with endometrial aspiration, is recommended beginning between ages 25 and 35 years because the increased risk for gynecologic cancer in these patients begins at age 25.

^67,68

There are no data demonstrating the efficacy of this type of screening. One recent study has even shown such screening not to be effective.

^69,70

It is important to note that, should a patient from an HNPCC family be diagnosed with endometrial cancer, they should undergo surveillance colonoscopy before hysterectomy in the event that colonic pathology is present and colonic resection is required at the same surgery.

Treatment

As with other colorectal conditions, there is not one operation that suits all patients. The majority of patients with a diagno- sis of HNPCC, based on the finding of an existing colorectal cancer or an advanced adenoma, will be offered colectomy and ileorectal anastomosis (IRA). This is thought to be the operation of choice, which removes the “at risk” organ, reduces cancer risk, preserves anal sphincter function, and retains the reservoir capacity of the rectum. This operation eliminates the need for annual surveillance colonoscopy, because only rigid or fiberoptic examination of the rectum is required. The estimated risk of rectal cancer after colectomy

and IRA is, however, 12% at 12 years.

⁷¹

This operation is clearly not suited for those who will not be compliant with follow-up examinations. Colectomy and IRA may not be the ideal operation for the patient with impaired anal sphincter function because of either obstetric injury or age or for patients with decreased mobility. The operation also is not suited for patients with more than three bowel movements per day, because these individuals will frequently have significant diarrhea and poor functional results after colectomy. In some patients, a lesser resection may be in order as a result of any of these reasons. It is however essential that both the patient and physician recognize the need for ongoing annual colonoscopy, because the risk for a metachronous colon can- cer in HNPCC is 45%.

⁷²Conversely, in the very young patient

with a long projected life expectancy or in the patient who is not likely to be compliant with follow-up examinations, total colectomy with stapled ileal pouch anal anastomosis may be the preferred procedure, because it removes nearly all of the at-risk colorectal mucosa.

In cases of rectal cancer not involving the sphincters, either an anterior resection, coloanal anastomosis or colectomy and ileal pouch anal anastomosis can be considered. In the event that the latter option is chosen, preoperative endorectal ultrasound stag- ing is desirable. In cases of a uT2 or uT3 or possible uN1 can- cers, preoperative chemoradiation should be given whenever possible, because ileal pouches tolerate radiation poorly.

In women undergoing colectomy, strong consideration should be given to performing a hysterectomy and bilateral salpingo-oophorectomy in the patient who has completed childbearing, because of the increased risk of both endome- trial and ovarian carcinoma.

There is controversy as to whether surgical treatment or continued surveillance should be offered to the asymptomatic patient who has a mutation identified by genetic testing, but an as yet “normal” colon. Several studies have examined this question using decision analysis methods; however, factors such as patient compliance must be taken into account.

^73,74

Prognosis

The survival rate in HNPCC patients with colorectal cancer is

better than that of patients with sporadic colorectal cancer

when matched for stage and age of onset.

^75,76

There is some

question of whether or not patients with Stage II or III

microsatellite unstable colorectal cancers benefit from 5-fluo-

rouracil-based adjuvant therapy. Whereas some studies report

that patients with microsatellite unstable tumors do not bene-

fit from such therapy, other studies suggest a benefit.

^77,78

One

possible explanation for the differences in such studies could

be attributable to different underlying causes of this MSI

(e.g., MLH1 gene-promoter methylation in some patients with

sporadic colorectal cancer). Although germline mutations are

involved in HNPCC, other mechanisms such as somatic

(10)

mutations and the loss of heterozygosity of mismatch-repair genes may be involved in microsatellite unstable tumors.

⁷⁹

Chemoprevention

Currently, there is much interest in chemoprevention. Whereas data exist regarding the efficacy of nonsteroidal antiinflamma- tory drugs (NSAIDs) in reducing the risk of colorectal cancer in the general population, such data are lacking for HNPCC.

^80,81

Currently, NSAIDs and novelose undigested starch are being studied in HNPCC. The CAPP II trial is a controlled, random- ized trial of colorectal polyp and cancer prevention using aspirin and resistant starch in carriers of HNPCC. It is a Phase II multi- center study with a multifactorial design, being performed at 33 centers. Patients are randomized to active agent or placebo, the active agents consisting of 600 mg of aspirin and 30 g of resist- ant starch for 2–4 years.

⁸²

The accrual goal is 1000 HNPCC gene carriers. This study, which began in January 1999 and is projected to end in December 2006, is designed to detect differ- ences in adenoma incidence.

^69,82

Calcium and vitamin D intake have been associated with a decreased risk of sporadic colorectal cancer.

⁸³

Supplemental dietary calcium is thought to inhibit the proliferative effect of bile acids and fatty acids on the colonic mucosa by precipitat- ing these luminal surfactants.

⁸⁴

A trial of supplemental calcium in HNPCC families did not demonstrate a decrease in epithelial proliferation, although the sample size was small and the study was conducted before genetic testing was available.

^69,85

Conclusion

Our knowledge about HNPCC continues to grow. This disorder is associated with a germline mutation in one of several MMR genes. Genetic testing is currently available for mutations in genes hMLH1 and hMHS2. Suspicion of this disorder in a given patient is raised by a family history of early age onset cancer, an increased number of first-degree relatives with colorectal cancer, or an HNPCC-related cancer. Endometrial cancer is the most common extracolonic cancer. Although there is debate regarding the frequency and beginning of screening, most cli- nicians believe that colonic examinations yearly or every 2 years should be performed beginning in patients in their early 20s. Prophylactic colectomy may be offered to known mutation carriers as well as members of at-risk families who develop advanced adenomas. Removal of all of the colon and IRA or colectomy and IPAA are the operations of choice, because they remove most “at-risk” colonic mucosa. In the event of the for- mer, continued surveillance of the rectum is mandatory. With careful surveillance and management, colorectal cancer can be prevented in these patients and the mortality rate decreased. In those who develop colorectal carcinoma, the overall survival rate is more favorable than that of patients with sporadic colorectal cancer.

Appendix I: Practice Parameters for the Identification and Testing of Patients at Risk for Dominantly Inherited Colorectal Cancer

Prepared by The Standards Task Force, The American Society of Colon and Rectal Surgeons, The Collaborative Group of the Americas on Inherited Colorectal Cancer

–Take a family history. This is the first step in recognizing families possibly affected by inherited colorectal cancer.

–Document a suspicious pedigree; a family tree based on the recollection of family members is not solid enough evidence. Request medical records to confirm diagnosis.

–Identify criteria for genetic testing. FAP is easily recog- nized clinically when patients present with more than 100 col- orectal adenomas. Fewer adenomas are needed for a diagnosis when a patient is part of an established kindred. The Amsterdam criteria are a way of clinically identifying fami- lies with hereditary nonpolyposis colorectal cancer, where an MMR gene mutation can be detected.

–Testing for MSI in tumors is a screen for families with hereditary nonpolyposis colorectal cancer where the clinical pattern of the disease is suggestive but not strong enough to fulfill Amsterdam criteria.

–Offer surveillance to families not meeting the above criteria for genetic testing. Families with more than two first-degree relatives affected with colorectal cancer, especially if one is affected at a young age (< 45 years), need to be offered endo- scopic surveillance even if genetic testing is not indicated.

–Adhere to all protocols for genetic testing. Institutional review board approval, informed consent, and pretest and posttest counseling are the key elements of genetic testing for inherited colorectal cancer.

Summary: It is hoped that these guidelines will assist in the recognition and management of patients affected by syn- dromes of inherited colorectal cancer.

Reprinted from Dis Colon Rectum 2001;44(10):1403–1412.

Appendix II: Practice Parameters for the Treatment of Patients with Dominantly Inherited Colorectal Cancer (FAP and Hereditary Nonpolyposis Colorectal Cancer)

Prepared by The Standards Task Force, The American Society of Colon and Rectal Surgeons

James Church, MD; Clifford Simmang, MD; on behalf of

the Collaborative Group of the Americas on Inherited

Colorectal Cancer and the Standards Committee of The

American Society of Colon and Rectal Surgeons.

(11)

Section 1. Familial Adenomatous Polyposis Guideline 1: Treatment Must Be Preceded by Thorough Counseling about the Nature of the Syndrome, Its Natural History, Its Extracolonic Manifestations, and the Need for Compliance with Recommendations for Management and Surveillance; Level of Evidence: III

Dominantly inherited colorectal cancer syndromes show a striking pattern of cancer in affected families. This is because of the high penetrance (penetrance

=

percent of patients with the mutation who have the disease) and often-severe expression (expression

=

clinical consequences of the mutation) of the mutations involved. FAP has a penetrance of close to 100%, colorectal cancer occurs at an average age of 39 years, and every affected patient is guaranteed at least one major abdomi- nal surgery. Despite these calamitous prospects, families with FAP adapt well to their disease. Most patients are compliant with recommended treatments, take a keen interest in the syn- drome, and have an active role in encouraging relatives to undergo screening. However, when a relative has a bad out- come, either because of severe disease or complications of treatment, family psychology may be affected. Noncompliance, denial, or a refusal to accept recommendations may ensue. The best way of avoiding both bad outcomes and an unfortunate response to them is to provide comprehensive, integrated coun- seling, support, and clinical services. These sorts of services are best provided through a department, registry, or center with personnel who have experience in managing patients and fam- ilies with these syndromes.

Guideline 2: Prophylactic Colectomy or Proctocolectomy Is Routine. The Timing and Type of Surgery Depend on the Severity of the Polyposis Phenotype and to a Lesser Extent on the Genotype, Age, and Clinical and Social Circumstances of the Patient;

Level of Evidence: III

The recommendation for prophylactic colectomy or procto- colectomy in FAP is based on the very high rates of colorectal cancer seen in patients who are not screened. In unscreened patients, the incidence of cancer is more than 60%.

Appropriate screening and timely surgery can minimize this.

The risk of cancer is not uniform, however, and is related to the severity of the colonic polyposis. Debinski and coworkers showed the rate of cancer for patients with >1000 colonic polyps was twice that of patients with <1000 colonic polyps.

In its turn, the severity of the colorectal polyposis is often related to the site of the APC mutation in a family. The “hot spot” mutation at codon 1309 is in an area of the gene where mutations always cause severe disease. Mutations in codons 3 and 4 are classically associated with attenuated FAP whereas mutations in the part of codon 15 that is 3

′

of codon 1450 are

usually associated with mild colorectal disease. Mutations in exons 5–15E have a variable colorectal phenotype, where some family members have relatively mild disease and others severe. The important aspects of surgery to consider are its timing, its type, and the technical options to be used.

Timing of Surgery

Even in patients with severe disease, cancer is rare under the age of 20. At-risk family members start screening (either genetic or with flexible sigmoidoscopy) at around puberty. If there is a positive genotype or an adenoma is seen on sig- moidoscopy, colonoscopy is recommended. The risk of can- cer of any individual patient can be estimated from the size and number of the adenomas seen on colonoscopy, and sur- gery planned accordingly. For patients with mild disease and low cancer risk, surgery can be done in mid-teen years (15–18 years). Where there is severe disease, or if the patient is symptomatic, surgery is done as soon as convenient after diagnosis.

Type of Surgery

There are three main surgical options: colectomy and IRA, proctocolectomy with ileostomy (TPC), and proctocolectomy with ileal pouch–anal anastomosis (IPAA). For any of these options, there are choices of technique. The IPAA can be sta- pled, leaving 1–2 cm of anal transitional epithelium and low rectal mucosa, or it can be hand-sewn after a complete anal mucosectomy. The operation can be done conventionally (i.e., open), laparoscopically, or laparoscopically assisted. The ileostomy may be a regular end stoma or one of the varieties of continent ileostomy (K or T).

Choice of Procedure

TPC is almost never done as a first operation except when a proctocolectomy is required and there is a contraindication to a pouch-anal anastomosis (e.g., a mesenteric desmoid tumor prevents the pouch from reaching the pelvic floor, a low rec- tal cancer invades the pelvic floor, or poor sphincters mean inability to control stool).

There is debate among authorities on which of the other

two options should be preferred. Some recommend IPAA

for all or almost all FAP patients, basing their recommenda-

tion on the risk of rectal cancer after IRA and equivalent

quality of life after the two operations. Others have shown

better functional outcomes after IRA and recommend it for

patients with mild colorectal polyposis. However, the risk

estimates of rectal cancer that are an overriding concern for

the proponents of universal IPAA are based on data col-

lected before restorative proctocolectomy was an option and

may well be overestimates, especially when applied to

patients with mild disease. The risk of rectal cancer after

IRA is strongly related to the severity of colorectal polypo-

sis at presentation, and IRA is a reasonable option in mildly

(12)

affected patients (< 20 rectal adenomas, < 1000 colonic adenomas). Retrospective data show that such patients have a very low risk of rectal cancer and include all those with attenuated FAP.

³⁰

Bowel function is usually good after IRA, the operation is simple, and complication rates are relatively low. Bowel function after a stapled IPAA is almost as good as with an IRA, and the anastomosis is usually safe enough to allow consideration of the option of avoiding a temporary ileostomy.

There is no argument that patients with severe rectal (> 20 adenomas) or colonic (>1000 adenomas), or those with a severely dysplastic rectal adenoma, a cancer anywhere in the large bowel, or a large (> 3 cm) rectal adenoma should have a primary IPAA. A stapled IPAA is associated with a risk of anal transitional neoplasia in 30% of patients, although if seri- ous neoplasia occurs (high-grade dysplasia or carpeting of the mucosa), the transitional zone can usually be stripped transanally and the pouch advanced to the dentate line. Even mucosectomy and hand-sewn IPAA is associated with anal neoplasia, although at a lower rate. The disadvantage of anal mucosectomy is worse function and increased complication rates. Both IRA and IPAA require lifelong surveillance of the rectum or pouch, because both are at risk of developing adenomas.

Choice of Technique

Mobilization of the colon using minimally invasive tech- niques such as laparoscopy or a Pfannenstiel incision is ideal for performing colectomy in children, because it minimizes the trauma of the surgery and the pain of the incisions. Its cos- metic result is appealing and it allows an early return to full activities. Whether minimally invasive techniques lower the risk of postoperative intraabdominal desmoid tumors is unknown, but the concept is attractive. A preoperative erect abdominal X-ray will usually show the position of the flex- ures and indicate whether use of a Pfannenstiel incision for mobilizing the colon is feasible.

Guideline 3: Lifetime Follow-up of the Rectum (after IRA), Pouch (after IPAA), and Ileostomy (after TPC) Is Required; Increasing Neoplasia in the Rectum Is an Indication for Proctectomy;

Level of Evidence: III

The combination of a germline APC mutation, stasis of stool, and glandular epithelium is potent at producing epithelial neoplasia. Adenomas and carcinomas have been described in the rectum, the ileostomy, and the ileal pouch itself, with the risk and severity of neoplasia increasing with time. The risk of severe neoplasia is mainly determined by the position of the mutation in the gene, as reflected by the severity of the polyposis. Severely affected patients have such a high risk of rectal cancer after IRA that subsequent proctectomy is

almost routine and initial IPAA is to be preferred. Yearly endoscopic surveillance of the bowel after the index surgery for FAP is standard. Two-thirds of patients undergo sponta- neous regression of rectal polyps after IRA, an effect that lasts 3–4 years. Subsequent surveillance will give a picture of the stability of the rectal mucosa. Small (< 5 mm) adenomas can be watched, although random biopsies are done to exclude severe dysplasia. Increasing number and size of ade- nomas are indications for more frequent surveillance, and adenomas > 5 mm should be removed cleanly with a snare.

Repeated fulguration of rectal polyps over many years can cause dense scarring that makes cancers flat and hard to see, and rectal dissection during proctectomy can be very difficult. Chronic rectal scarring makes rectal biopsy diffi- cult, because the forceps tend to “bounce off” the scarred mucosa. Furthermore, scarring leads to reduced rectal com- pliance, increased stool frequency, and a tendency to seepage and incontinence. Severe dysplasia, or villous adenomas >1 cm, are indications for proctectomy. Proctoscopy is best done with a video endoscope, because comfort is enhanced and the view is better. Excellent preparation and a good view are essential to pick up early cancers that can be flat and subtle.

Sulindac (Clinoril®; Merck & Co., Inc., West Point, PA), either by mouth or by suppository, is effective in making polyps disappear. Celecoxib reduces polyp load, as does the sulindac metabolite exisulind. However, cancers have been reported in cases where sulindac had been effective in mini- mizing rectal polyps in the rectum of FAP patients who had had IRA, and these anecdotal cases make the long-term use of chemoprevention for rectal polyposis suspect. If it is used in patients who cannot tolerate rectal polypectomy, or who are unwilling or unable to have a proctectomy, close surveillance (every 6 months) with random biopsies to look for severe dysplasia is needed.

There have been at least three recent reports describing adenomas in ileal pouches, with a frequency and severity that depend on time from the initial surgery. Two prospective studies have independently calculated the rate of pouch poly- posis as 42% at 7 years. There have been anecdotal reports of large adenomas and more than 100 adenomas in an ileal pouch. In general, these have been treated successfully by oral sulindac, in a dose of 150–200 mg twice daily. The full impact of pouch polyposis will not be obvious until the cadre of FAP patients with ileal reservoirs reaches a mean follow- up of 20 years. This is the time to most ileostomy cancers, and to the highest rates of rectal cancers after IRA.

Guideline 4: Use of Chemoprevention as Primary Therapy for Colorectal Polyposis Is Not Proven and Is Not Recommended;

Level of Evidence: I–II

Sulindac, celecoxib, and exisulind are nonsteroidal antiin-

flammatory drugs that have been shown to reduce the num-

(13)

ber and size of colorectal adenomas in patients with FAP.

Although many studies are short-term, two show effective- ness of sulindac maintained over 4 years. These studies were in patients who had undergone colectomy and IRA.

A recent randomized, placebo-controlled, double-blind study of sulindac in genotype-positive, phenotype-negative FAP patients failed to show any effect of sulindac on polyp progression. Furthermore, there have been case reports of cancers occurring in patients with sulindac-mediated abla- tion of polyps, and the only report of a permanent, complete resolution of rectal polyposis comes from Winde and coworkers who used sulindac suppositories. The effect on polyps is dependent on continued compliance, and there are significant side effects with each medication. These med- ications should not be used as an alternative to surgery, except in patients with pouch polyposis or in selected patients with rectal polyposis after IRA in whom surgery is risky or unwanted by the patient. In these groups of patients, close surveillance (proctoscopy or pouchoscopy every 6 months) is indicated.

Guideline 5: Treatment of Duodenal Adenomas Depends on Adenoma Size and the Presence of Severe Dysplasia. Small Tubular Adenomas with Mild Dysplasia Can Be Kept under Surveillance, but Adenomas with Severe Dysplasia Must Be Removed; Level of Evidence: II–III

The incidence of duodenal adenomas in FAP patients is in the range of 80%–90%. All FAP patients therefore undergo screening esophagogastroduodenoscopy starting at age 20 years. The risk of invasive cancer developing in a duode- nal adenoma, or in the duodenal papilla, is considerably higher than that for the average population, but in absolute terms it is still low. The aim of endoscopy is not to eradicate all neoplasia but to make sure that there is no severe dyspla- sia. Studies of the natural history of duodenal neoplasia in FAP show that rapid progression of dysplasia is uncommon, occurring in only 11% of cases over a mean follow-up of 7 years. Prospective, randomized studies have shown that treatment with nonsteroidal antiinflammatory drugs is inef- fective in treating duodenal adenomas, although a recent report indicates that celecoxib may have some effect. If they are not medically treated, low-risk adenomas (small, tubular, low-grade dysplasia) may be biopsied and left alone. High- risk adenomas (>1 cm, villous) are treated. Adenomas with confirmed high-grade dysplasia must be removed. As endo- scopic or even transduodenal excision or destruction is inef- fective in the long term; duodenectomy has to be considered for duodenal adenomas with high-grade dysplasia after the diagnosis has been confirmed on review by an experienced gastrointestinal pathologist.

Guideline 6: Duodenectomy or

Pancreaticoduodenectomy Is Recommended for Patients with Persistent or Recurrent Severe Dysplasia in the Papilla or Duodenal Adenomas; Level of Evidence: III

A review of literature reporting treatment of advanced duode- nal adenomas shows that recurrence is almost guaranteed unless the duodenum is removed. Transduodenal polypec- tomy or endoscopic polypectomy may be temporarily effective, but does not offer a permanent cure. The results of pancreas-preserving duodenectomy or pancreaticoduodenec- tomy for benign or early malignant disease are good, with low recurrence and acceptable morbidity. The outcome of surgery for established cancer is not good, with recurrence and death the usual outcome. Although the risk of duodenal/peri- ampullary cancer is relatively low in patients with FAP, patients with persistent high-grade dysplasia in the duodenum or papilla are a high-risk group. Careful surveillance is needed, and conservative surgery or endoscopic therapy may be tried. If the severe dysplasia returns or persists, considera- tion must be given to duodenectomy.

Guideline 7: Surgery for Intraabdominal Desmoid Tumors Should Be Reserved for Small, Well-defined Tumors with a Clear Margin;

Abdominal Wall Desmoid Tumors Should Be Excised Whenever Possible; Level of Evidence: III

Desmoid tumors are histologically benign overgrowths of fibroaponeurotic tissue occurring rarely in the general popu- lation but in 12%–17% of patients with FAP. In the general population, desmoids are usually found in limbs or limb gir- dles; in FAP, desmoids are usually (80%) intraabdominal and often (80%) present within 2–3 years of an abdominal sur- gery. Intraabdominal desmoid tumors usually involve the mesentery of the small bowel, where they are intimately involved with the mesenteric vessels. They tend to infiltrate diffusely, kink adjacent bowel loops, and may obstruct the ureters. Attempts at excision are often unsuccessful, involve removal of a variable length of small intestine, and are associated with a high morbidity and a high recurrence.